Merge pull request #18436 from github/release-prep/2.20.1

Release preparation for version 2.20.1

.bazelrc

@@ -1,5 +1,4 @@
 common --enable_platform_specific_config
-common --enable_bzlmod
 # because we use --override_module with `%workspace%`, the lock file is not stable
 common --lockfile_mode=off
 
@@ -24,6 +23,13 @@ common --registry=file:///%workspace%/misc/bazel/registry
 common --registry=https://bcr.bazel.build
 
 common --@rules_dotnet//dotnet/settings:strict_deps=false
-common --experimental_isolated_extension_usages
+
+# Reduce this eventually to empty, once we've fixed all our usages of java, and https://github.com/bazel-contrib/rules_go/issues/4193 is fixed
+common --incompatible_autoload_externally="+@rules_java,+@rules_shell"
+
+build --java_language_version=17
+build --tool_java_language_version=17
+build --tool_java_runtime_version=remotejdk_17
+build --java_runtime_version=remotejdk_17
 
 try-import %workspace%/local.bazelrc

.bazelrc.internal

@@ -8,4 +8,3 @@ common --registry=https://bcr.bazel.build
 # its implementation packages without providing any code itself.
 # We either can depend on internal implementation details, or turn of strict deps.
 common --@rules_dotnet//dotnet/settings:strict_deps=false
-common --experimental_isolated_extension_usages

.bazelversion

@@ -1 +1 @@
-5f5d70b6c4d2fb1a889479569107f1692239e8a7
+8.0.0

.devcontainer/devcontainer.json

@@ -1,4 +1,5 @@
 {
+  "image": "mcr.microsoft.com/devcontainers/base:ubuntu-24.04",
   "extensions": [
     "rust-lang.rust-analyzer",
     "bungcip.better-toml",

.gitattributes

@@ -86,4 +86,5 @@
 /misc/ripunzip/ripunzip-* filter=lfs diff=lfs merge=lfs -text
 
 # swift prebuilt resources
-/swift/third_party/resource-dir/*.zip filter=lfs diff=lfs merge=lfs -text
+/swift/third_party/resources/*.zip filter=lfs diff=lfs merge=lfs -text
+/swift/third_party/resources/*.tar.zst filter=lfs diff=lfs merge=lfs -text

.github/codeql/codeql-config.yml

@@ -9,3 +9,4 @@ paths-ignore:
   - '/python/'
   - '/javascript/ql/test'
   - '/javascript/extractor/tests'
+  - '/rust/ql'

.github/labeler.yml

@@ -38,6 +38,10 @@ Swift:
   - swift/**/*
   - change-notes/**/*swift*
 
+Actions:
+  - actions/**/*
+  - change-notes/**/*actions*
+
 documentation:
   - "**/*.qhelp"
   - "**/*.md"

.github/workflows/build-ripunzip.yml

@@ -17,7 +17,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04, macos-12, windows-2019]
+        os: [ubuntu-20.04, macos-13, windows-2019]
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v4

.github/workflows/check-qldoc.yml

@@ -30,7 +30,8 @@ jobs:
         run: |
           EXIT_CODE=0
           # TODO: remove the shared exception from the regex when coverage of qlpacks without dbschemes is supported
-          changed_lib_packs="$(git diff --name-only --diff-filter=ACMRT HEAD^ HEAD | { grep -Po '^(?!(shared))[a-z]*/ql/lib' || true; } | sort -u)"
+          # TODO: remove the actions exception once https://github.com/github/codeql-team/issues/3656 is fixed
+          changed_lib_packs="$(git diff --name-only --diff-filter=ACMRT HEAD^ HEAD | { grep -Po '^(?!(shared|actions))[a-z]*/ql/lib' || true; } | sort -u)"
           for pack_dir in ${changed_lib_packs}; do
             lang="${pack_dir%/ql/lib}"
             codeql generate library-doc-coverage --output="${RUNNER_TEMP}/${lang}-current.txt" --dir="${pack_dir}"

.github/workflows/codeql-analysis.yml

@@ -30,7 +30,7 @@ jobs:
     - name: Setup dotnet
       uses: actions/setup-dotnet@v4
       with:
-        dotnet-version: 8.0.101
+        dotnet-version: 9.0.100
 
     - name: Checkout repository
       uses: actions/checkout@v4

.github/workflows/compile-queries.yml

@@ -7,6 +7,11 @@ on:
       - "rc/*"
       - "codeql-cli-*"
   pull_request:
+    paths:
+      - '**.ql'
+      - '**.qll'
+      - '**/qlpack.yml'
+      - '**.dbscheme'
 
 permissions:
   contents: read
@@ -33,9 +38,9 @@ jobs:
         # run with --check-only if running in a PR (github.sha != main)
         if : ${{ github.event_name == 'pull_request' }}
         shell: bash
-        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --check-only --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500
+        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --check-only --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500 --ram=56000
       - name: compile queries - full
         # do full compile if running on main - this populates the cache
         if : ${{ github.event_name != 'pull_request' }}
         shell: bash
-        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500
+        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500 --ram=56000

.github/workflows/cpp-swift-analysis.yml

@@ -19,7 +19,7 @@ on:
 jobs:
   CodeQL-Build:
 
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
 
     permissions:
       contents: read
@@ -38,17 +38,15 @@ jobs:
         languages: cpp
         config-file: ./.github/codeql/codeql-config.yml
 
-    - name: "[Ubuntu] Remove GCC 13 from runner image"
-      shell: bash
+    - name: Install dependencies
       run: |
-        sudo rm -f /etc/apt/sources.list.d/ubuntu-toolchain-r-ubuntu-test-jammy.list
         sudo apt-get update
-        sudo apt-get install -y --allow-downgrades libc6=2.35-* libc6-dev=2.35-* libstdc++6=12.3.0-* libgcc-s1=12.3.0-*
+        sudo apt-get install -y uuid-dev
 
     - name: "Build Swift extractor using Bazel"
       run: |
         bazel clean --expunge
-        bazel run //swift:create-extractor-pack --nouse_action_cache --noremote_accept_cached --noremote_upload_local_results --spawn_strategy=local
+        bazel run //swift:install --nouse_action_cache --noremote_accept_cached --noremote_upload_local_results --spawn_strategy=local
         bazel shutdown
 
     - name: Perform CodeQL Analysis

.github/workflows/csharp-qltest.yml

@@ -5,8 +5,10 @@ on:
     paths:
       - "csharp/**"
       - "shared/**"
+      - "misc/bazel/**"
       - .github/actions/fetch-codeql/action.yml
       - codeql-workspace.yml
+      - "MODULE.bazel"
     branches:
       - main
       - "rc/*"
@@ -14,9 +16,11 @@ on:
     paths:
       - "csharp/**"
       - "shared/**"
+      - "misc/bazel/**"
       - .github/workflows/csharp-qltest.yml
       - .github/actions/fetch-codeql/action.yml
       - codeql-workspace.yml
+      - "MODULE.bazel"
     branches:
       - main
       - "rc/*"
@@ -39,14 +43,14 @@ jobs:
       - name: Setup dotnet
         uses: actions/setup-dotnet@v4
         with:
-          dotnet-version: 8.0.101
+          dotnet-version: 9.0.100
       - name: Extractor unit tests
         run: |
           dotnet tool restore
-          dotnet test -p:RuntimeFrameworkVersion=8.0.1 extractor/Semmle.Util.Tests
-          dotnet test -p:RuntimeFrameworkVersion=8.0.1 extractor/Semmle.Extraction.Tests
-          dotnet test -p:RuntimeFrameworkVersion=8.0.1 autobuilder/Semmle.Autobuild.CSharp.Tests
-          dotnet test -p:RuntimeFrameworkVersion=8.0.1 autobuilder/Semmle.Autobuild.Cpp.Tests
+          dotnet test -p:RuntimeFrameworkVersion=9.0.0 extractor/Semmle.Util.Tests
+          dotnet test -p:RuntimeFrameworkVersion=9.0.0 extractor/Semmle.Extraction.Tests
+          dotnet test -p:RuntimeFrameworkVersion=9.0.0 autobuilder/Semmle.Autobuild.CSharp.Tests
+          dotnet test -p:RuntimeFrameworkVersion=9.0.0 autobuilder/Semmle.Autobuild.Cpp.Tests
         shell: bash
   stubgentest:
     runs-on: ubuntu-latest

.github/workflows/csv-coverage-metrics.yml

@@ -37,7 +37,7 @@ jobs:
         run: |
           DATABASE="${{ runner.temp }}/java-database"
           codeql database analyze --format=sarif-latest --output=metrics-java.sarif -- "$DATABASE" ./java/ql/src/Metrics/Summaries/FrameworkCoverage.ql
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: metrics-java.sarif
           path: metrics-java.sarif
@@ -64,7 +64,7 @@ jobs:
         run: |
           DATABASE="${{ runner.temp }}/csharp-database"
           codeql database analyze --format=sarif-latest --output=metrics-csharp.sarif -- "$DATABASE" ./csharp/ql/src/Metrics/Summaries/FrameworkCoverage.ql
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: metrics-csharp.sarif
           path: metrics-csharp.sarif

.github/workflows/csv-coverage-pr-artifacts.yml

@@ -71,21 +71,21 @@ jobs:
         run: |
           python base/misc/scripts/library-coverage/compare-folders.py out_base out_merge comparison.md
       - name: Upload CSV package list
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: csv-framework-coverage-merge
           path: |
             out_merge/framework-coverage-*.csv
             out_merge/framework-coverage-*.rst
       - name: Upload CSV package list
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: csv-framework-coverage-base
           path: |
             out_base/framework-coverage-*.csv
             out_base/framework-coverage-*.rst
       - name: Upload comparison results
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: comparison
           path: |
@@ -97,7 +97,7 @@ jobs:
         env:
           PR_NUMBER: ${{ github.event.pull_request.number }}
       - name: Upload PR number
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: pr
           path: pr/
@@ -117,7 +117,7 @@ jobs:
           GITHUB_TOKEN: ${{ github.token }}
           PR_NUMBER: ${{ github.event.pull_request.number }}
       - name: Upload comment ID (if it exists)
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: comment
           path: comment/

.github/workflows/csv-coverage-timeseries.yml

@@ -30,7 +30,7 @@ jobs:
         run: |
           python script/misc/scripts/library-coverage/generate-timeseries.py codeqlModels
       - name: Upload timeseries CSV
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: framework-coverage-timeseries
           path: framework-coverage-timeseries-*.csv

.github/workflows/csv-coverage.yml

@@ -34,12 +34,12 @@ jobs:
         run: |
           python script/misc/scripts/library-coverage/generate-report.py ci codeqlModels script
       - name: Upload CSV package list
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: framework-coverage-csv
           path: framework-coverage-*.csv
       - name: Upload RST package list
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: framework-coverage-rst
           path: framework-coverage-*.rst

.github/workflows/mad_modelDiff.yml

@@ -38,14 +38,20 @@ jobs:
           path: codeql-main
           ref: main
       - uses: ./codeql-main/.github/actions/fetch-codeql
+      # compute the shortname of the project that does not contain any special (disk) characters
+      - run: |
+          echo "SHORTNAME=${SLUG//[^a-zA-Z0-9_]/}" >> $GITHUB_OUTPUT
+        env: 
+          SLUG: ${{ matrix.slug }}
+        id: shortname
       - name: Download database
         env:
           SLUG: ${{ matrix.slug }}
           GH_TOKEN: ${{ github.token }}
+          SHORTNAME: ${{ steps.shortname.outputs.SHORTNAME }}
         run: |
           set -x
           mkdir lib-dbs
-          SHORTNAME=${SLUG//[^a-zA-Z0-9_]/}
           gh api -H "Accept: application/zip" "/repos/${SLUG}/code-scanning/codeql/databases/java" > "$SHORTNAME.zip"
           unzip -q -d "${SHORTNAME}-db" "${SHORTNAME}.zip"
           mkdir "lib-dbs/$SHORTNAME/"
@@ -93,14 +99,14 @@ jobs:
             name="diff_${basename/.model.yml/""}"
             (diff -w -u $m $t | diff2html -i stdin -F $MODELS/$name.html) || true
           done
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: models
+          name: models-${{ steps.shortname.outputs.SHORTNAME }}
           path: tmp-models/**/**/*.model.yml
           retention-days: 20
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: diffs
+          name: diffs-${{ steps.shortname.outputs.SHORTNAME }}
           path: tmp-models/*.html
           # An html file is only produced if the generated models differ.
           if-no-files-found: ignore

.github/workflows/mad_regenerate-models.yml

@@ -59,7 +59,7 @@ jobs:
           find java -name "*.model.yml" -print0 | xargs -0 git add
           git status
           git diff --cached > models.patch
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: patch
           path: models.patch

.github/workflows/post-pr-comment.yml

@@ -17,8 +17,11 @@ jobs:
   post_comment:
     runs-on: ubuntu-latest
     steps:
-      - name: Download artifact
-        run: gh run download "${WORKFLOW_RUN_ID}" --repo "${GITHUB_REPOSITORY}" --name "comment"
+      - name: Download artifacts
+        run: |
+          gh run download "${WORKFLOW_RUN_ID}" --repo "${GITHUB_REPOSITORY}" --name "comment-pr-number"
+          gh run download "${WORKFLOW_RUN_ID}" --repo "${GITHUB_REPOSITORY}" --name "comment-body"
+          gh run download "${WORKFLOW_RUN_ID}" --repo "${GITHUB_REPOSITORY}" --name "comment-id"
         env:
           GITHUB_TOKEN: ${{ github.token }}
           WORKFLOW_RUN_ID: ${{ github.event.workflow_run.id }}

.github/workflows/qhelp-pr-preview.yml

@@ -36,9 +36,9 @@ jobs:
       - run: echo "${PR_NUMBER}" > pr_number.txt
         env:
           PR_NUMBER: ${{ github.event.number }}
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: comment
+          name: comment-pr-number
           path: pr_number.txt
           if-no-files-found: error
           retention-days: 1
@@ -78,9 +78,9 @@ jobs:
           exit "${EXIT_CODE}"
 
       - if: ${{ !cancelled() }}
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
-          name: comment
+          name: comment-body
           path: comment_body.txt
           if-no-files-found: error
           retention-days: 1
@@ -94,9 +94,9 @@ jobs:
           GITHUB_TOKEN: ${{ github.token }}
           PR_NUMBER: ${{ github.event.number }}
 
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: comment
+          name: comment-id
           path: comment_id.txt
           if-no-files-found: error
           retention-days: 1

.github/workflows/ql-for-ql-build.yml

@@ -75,7 +75,7 @@ jobs:
           sarif_file: ql-for-ql.sarif
           category: ql-for-ql
       - name: Sarif as artifact
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: ql-for-ql.sarif
           path: ql-for-ql.sarif
@@ -84,7 +84,7 @@ jobs:
           mkdir split-sarif
           node ./ql/scripts/split-sarif.js ql-for-ql.sarif split-sarif
       - name: Upload langs as artifacts
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: ql-for-ql-langs
           path: split-sarif

.github/workflows/ql-for-ql-dataset_measure.yml

@@ -65,7 +65,7 @@ jobs:
           "${CODEQL}" dataset measure --threads 4 --output "stats/${{ matrix.repo }}/stats.xml" "${{ runner.temp }}/database/db-ql"
         env:
           CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: measurements
           path: stats
@@ -76,14 +76,14 @@ jobs:
     needs: measure
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: measurements
           path: stats
       - run: |
           python -m pip install --user lxml
           find stats -name 'stats.xml' -print0 | sort -z | xargs -0 python ruby/scripts/merge_stats.py --output ql/ql/src/ql.dbscheme.stats --normalise ql_tokeninfo
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: ql.dbscheme.stats
           path: ql/ql/src/ql.dbscheme.stats

.github/workflows/ql-for-ql-tests.yml

@@ -40,7 +40,7 @@ jobs:
             ql/target
           key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-qltest-cargo-${{ hashFiles('ql/rust-toolchain.toml', 'ql/**/Cargo.lock') }}
       - name: Check formatting
-        run: cd ql; cargo fmt --all -- --check
+        run: cd ql; cargo fmt -- --check
       - name: Build extractor
         run: |
           cd ql;

.github/workflows/query-list.yml

@@ -37,7 +37,7 @@ jobs:
       run: |
         python codeql/misc/scripts/generate-code-scanning-query-list.py > code-scanning-query-list.csv
     - name: Upload code scanning query list
-      uses: actions/upload-artifact@v3
+      uses: actions/upload-artifact@v4
       with:
         name: code-scanning-query-list
         path: code-scanning-query-list.csv

.github/workflows/ruby-build.yml

@@ -79,7 +79,7 @@ jobs:
           key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-ruby-rust-cargo-${{ hashFiles('ruby/extractor/rust-toolchain.toml', 'ruby/extractor/**/Cargo.lock') }}
       - name: Check formatting
         if: steps.cache-extractor.outputs.cache-hit != 'true'
-        run: cd extractor && cargo fmt --all -- --check
+        run: cd extractor && cargo fmt -- --check
       - name: Build
         if: steps.cache-extractor.outputs.cache-hit != 'true'
         run: cd extractor && cargo build --verbose
@@ -92,17 +92,17 @@ jobs:
       - name: Generate dbscheme
         if: ${{ matrix.os == 'ubuntu-latest' && steps.cache-extractor.outputs.cache-hit != 'true'}}
         run: ../target/release/codeql-extractor-ruby generate --dbscheme ql/lib/ruby.dbscheme --library ql/lib/codeql/ruby/ast/internal/TreeSitter.qll
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         if: ${{ matrix.os == 'ubuntu-latest' }}
         with:
           name: ruby.dbscheme
           path: ruby/ql/lib/ruby.dbscheme
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         if: ${{ matrix.os == 'ubuntu-latest' }}
         with:
           name: TreeSitter.qll
           path: ruby/ql/lib/codeql/ruby/ast/internal/TreeSitter.qll
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: extractor-${{ matrix.os }}
           path: |
@@ -134,7 +134,7 @@ jobs:
           PACK_FOLDER=$(readlink -f "$PACKS"/codeql/ruby-queries/*)
           codeql generate query-help --format=sarifv2.1.0 --output="${PACK_FOLDER}/rules.sarif" ql/src
           (cd ql/src; find queries \( -name '*.qhelp' -o -name '*.rb' -o -name '*.erb' \) -exec bash -c 'mkdir -p "'"${PACK_FOLDER}"'/$(dirname "{}")"' \; -exec cp "{}" "${PACK_FOLDER}/{}" \;)
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: codeql-ruby-queries
           path: |
@@ -147,19 +147,19 @@ jobs:
     needs: [build, compile-queries]
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: ruby.dbscheme
           path: ruby/ruby
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: extractor-ubuntu-latest
           path: ruby/linux64
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: extractor-windows-latest
           path: ruby/win64
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: extractor-macos-latest
           path: ruby/osx64
@@ -172,13 +172,13 @@ jobs:
           cp win64/codeql-extractor-ruby.exe ruby/tools/win64/extractor.exe
           chmod +x ruby/tools/{linux64,osx64}/extractor
           zip -rq codeql-ruby.zip ruby
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: codeql-ruby-pack
           path: ruby/codeql-ruby.zip
           retention-days: 1
           include-hidden-files: true
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
           name: codeql-ruby-queries
           path: ruby/qlpacks
@@ -190,7 +190,7 @@ jobs:
             ]
           }' > .codeqlmanifest.json
           zip -rq codeql-ruby-bundle.zip .codeqlmanifest.json ruby qlpacks
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: codeql-ruby-bundle
           path: ruby/codeql-ruby-bundle.zip
@@ -214,7 +214,7 @@ jobs:
         uses: ./.github/actions/fetch-codeql
 
       - name: Download Ruby bundle
-        uses: actions/download-artifact@v3
+        uses: actions/download-artifact@v4
         with:
           name: codeql-ruby-bundle
           path: ${{ runner.temp }}

.github/workflows/ruby-dataset-measure.yml

@@ -52,9 +52,9 @@ jobs:
         run: |
           mkdir -p "stats/${{ matrix.repo }}"
           codeql dataset measure --threads 4 --output "stats/${{ matrix.repo }}/stats.xml" "${{ runner.temp }}/database/db-ruby"
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: measurements
+          name: measurements-${{ hashFiles('stats/**') }}
           path: stats
           retention-days: 1
 
@@ -63,14 +63,13 @@ jobs:
     needs: measure
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v4
         with:
-          name: measurements
           path: stats
       - run: |
           python -m pip install --user lxml
           find stats -name 'stats.xml' | sort | xargs python ruby/scripts/merge_stats.py --output ruby/ql/lib/ruby.dbscheme.stats --normalise ruby_tokeninfo
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: ruby.dbscheme.stats
           path: ruby/ql/lib/ruby.dbscheme.stats

.github/workflows/rust-analysis.yml

@@ -0,0 +1,64 @@
+name: "Code scanning - Rust"
+
+on:
+  push:
+    branches:
+      - main
+      - 'rc/*'
+  pull_request:
+    branches:
+      - main
+      - 'rc/*'
+    paths:
+      - '**/*.rs'
+      - '**/Cargo.toml'
+      - '.github/codeql/codeql-config.yml'
+      - '.github/workflows/rust-analysis.yml'
+  schedule:
+    - cron: '0 9 * * 1'
+
+env:
+  CODEQL_ENABLE_EXPERIMENTAL_FEATURES: "true"
+
+jobs:
+  analyze:
+    strategy:
+      matrix:
+        language: [ 'rust' ]
+
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      security-events: write
+      pull-requests: read
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Query latest nightly CodeQL bundle
+      shell: bash
+      id: codeql
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        REPO=dsp-testing/codeql-cli-nightlies
+        TAG=$(
+          gh release list -R $REPO -L1 --exclude-drafts --json tagName -q ".[] | .tagName"
+        )
+        echo "nightly_bundle=https://github.com/$REPO/releases/download/$TAG/codeql-bundle-linux64.tar.zst" \
+          | tee -a "$GITHUB_OUTPUT"
+
+    - name: Initialize CodeQL
+      uses: github/codeql-action/init@main
+      with:
+        tools: ${{ steps.codeql.outputs.nightly_bundle }}
+        languages: ${{ matrix.language }}
+        config-file: ./.github/codeql/codeql-config.yml
+
+    - name: Autobuild
+      uses: github/codeql-action/autobuild@main
+
+    - name: Perform CodeQL Analysis
+      uses: github/codeql-action/analyze@main

.github/workflows/rust.yml

@@ -23,26 +23,48 @@ permissions:
   contents: read
 
 jobs:
-  rust-code:
+  rust-ast-generator:
     runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: rust/ast-generator
     steps:
       - name: Checkout
         uses: actions/checkout@v4
+      - name: Inject sources
+        shell: bash
+        run: |
+          bazel run //rust/ast-generator:inject-sources
       - name: Format
-        working-directory: rust/extractor
         shell: bash
         run: |
           cargo fmt --check
       - name: Compilation
-        working-directory: rust/extractor
         shell: bash
         run: cargo check
       - name: Clippy
-        working-directory: rust/extractor
         shell: bash
         run: |
-          cargo clippy --fix
-          git diff --exit-code
+          cargo clippy --no-deps -- -D warnings
+  rust-code:
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: rust/extractor
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Format
+        shell: bash
+        run: |
+          cargo fmt --check
+      - name: Compilation
+        shell: bash
+        run: cargo check
+      - name: Clippy
+        shell: bash
+        run: |
+          cargo clippy --no-deps -- -D warnings
   rust-codegen:
     runs-on: ubuntu-latest
     steps:

.github/workflows/swift.yml

@@ -44,27 +44,20 @@ jobs:
   # without waiting for the macOS build
   build-and-test-macos:
     if: github.repository_owner == 'github'
-    runs-on: macos-12-xl
+    runs-on: macos-13-xlarge
     steps:
       - uses: actions/checkout@v4
       - uses: ./swift/actions/build-and-test
   build-and-test-linux:
     if: github.repository_owner == 'github'
-    runs-on: ubuntu-latest-xl
+    runs-on: ubuntu-22.04
     steps:
       - uses: actions/checkout@v4
       - uses: ./swift/actions/build-and-test
-  qltests-linux:
-    if: github.repository_owner == 'github'
-    needs: build-and-test-linux
-    runs-on: ubuntu-latest-xl
-    steps:
-      - uses: actions/checkout@v4
-      - uses: ./swift/actions/run-ql-tests
   qltests-macos:
     if: ${{ github.repository_owner == 'github' && github.event_name == 'pull_request' }}
     needs: build-and-test-macos
-    runs-on: macos-12-xl
+    runs-on: macos-13-xlarge
     steps:
       - uses: actions/checkout@v4
       - uses: ./swift/actions/run-ql-tests
@@ -98,7 +91,7 @@ jobs:
       - name: Generate C++ files
         run: |
           bazel run //swift/codegen:codegen -- --generate=trap,cpp --cpp-output=$PWD/generated-cpp-files
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: swift-generated-cpp-files
           path: generated-cpp-files/**
@@ -109,3 +102,10 @@ jobs:
       - uses: actions/checkout@v4
       - uses: ./.github/actions/fetch-codeql
       - uses: ./swift/actions/database-upgrade-scripts
+  check-no-override:
+    if : github.event_name == 'pull_request'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - shell: bash
+        run: bazel test //swift/... --test_tag_filters=override --test_output=errors

.github/workflows/tree-sitter-extractor-test.yml

@@ -32,17 +32,17 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - name: Check formatting
-        run: cargo fmt --all -- --check
+        run: cargo fmt -- --check
       - name: Run tests
         run: cargo test --verbose
   fmt:
-    runs-on: ubuntu-latest  
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - name: Check formatting
         run: cargo fmt --check
   clippy:
-    runs-on: ubuntu-latest  
+    runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - name: Run clippy

.gitignore

@@ -8,7 +8,7 @@
 
 # qltest projects and artifacts
 *.actual
-*/ql/test/**/*.testproj
+*/ql/test*/**/*.testproj
 */ql/test/**/go.sum
 
 # Visual studio temporaries, except a file used by QL4VS

.pre-commit-config.yaml

@@ -72,7 +72,7 @@ repos:
 
       - id: rust-codegen
         name: Run Rust checked in code generation
-        files: ^misc/codegen/|^rust/(schema.py$|codegen/|.*/generated/|ql/lib/(rust\.dbscheme$|codeql/rust/elements)|\.generated.list)
+        files: ^misc/codegen/|^rust/(prefix\.dbscheme|schema/|codegen/|.*/generated/|ql/lib/(rust\.dbscheme$|codeql/rust/elements)|\.generated.list)
         language: system
         entry: bazel run //rust/codegen -- --quiet
         pass_filenames: false

.vscode/settings.json

@@ -1,5 +1,6 @@
 {
   "omnisharp.autoStart": false,
   "cmake.sourceDirectory": "${workspaceFolder}/swift",
-  "cmake.buildDirectory": "${workspaceFolder}/bazel-cmake-build"
+  "cmake.buildDirectory": "${workspaceFolder}/bazel-cmake-build",
+  "editor.suggest.matchOnWordStartOnly": false
 }

.vscode/tasks.json

@@ -38,6 +38,94 @@
                 "command": "${config:python.pythonPath}",
             },
             "problemMatcher": []
+        },
+        {
+            "label": "Create query change note",
+            "type": "process",
+            "command": "python3",
+            "args": [
+                "misc/scripts/create-change-note.py",
+                "${input:language}",
+                "src",
+                "${input:name}",
+                "${input:categoryQuery}"
+            ],
+            "presentation": {
+                "reveal": "never",
+                "close": true
+            },
+            "problemMatcher": []
+        },
+                {
+            "label": "Create library change note",
+            "type": "process",
+            "command": "python3",
+            "args": [
+                "misc/scripts/create-change-note.py",
+                "${input:language}",
+                "lib",
+                "${input:name}",
+                "${input:categoryLibrary}"
+            ],
+            "presentation": {
+                "reveal": "never",
+                "close": true
+            },
+            "problemMatcher": []
+        }
+    ],
+    "inputs": [
+        {
+            "type": "pickString",
+            "id": "language",
+            "description": "Language",
+            "options":
+            [
+                "actions",
+                "go",
+                "java",
+                "javascript",
+                "cpp",
+                "csharp",
+                "python",
+                "ruby",
+                "rust",
+                "swift",
+            ]
+        },
+        {
+            "type": "promptString",
+            "id": "name",
+            "description": "Short name (kebab-case)"
+        },
+        {
+            "type": "pickString",
+            "id": "categoryQuery",
+            "description": "Category (query change)",
+            "options":
+            [
+                "breaking",
+                "deprecated",
+                "newQuery",
+                "queryMetadata",
+                "majorAnalysis",
+                "minorAnalysis",
+                "fix",
+            ]
+        },
+                {
+            "type": "pickString",
+            "id": "categoryLibrary",
+            "description": "Category (library change)",
+            "options":
+            [
+                "breaking",
+                "deprecated",
+                "feature",
+                "majorAnalysis",
+                "minorAnalysis",
+                "fix",
+            ]
         }
     ]
 }

2024-11-25-ts57.md

@@ -0,0 +1,4 @@
+---
+category: majorAnalysis
+---
+* Added support for TypeScript 5.7.

BUILD.bazel

@@ -1 +1,5 @@
-exports_files(["LICENSE"])
+exports_files([
+    "LICENSE",
+    "Cargo.lock",
+    "Cargo.toml",
+])

CODEOWNERS

@@ -1,3 +1,4 @@
+/actions/ @github/codeql-dynamic
 /cpp/ @github/codeql-c-analysis
 /csharp/ @github/codeql-csharp
 /csharp/autobuilder/Semmle.Autobuild.Cpp @github/codeql-c-extractor
@@ -23,7 +24,6 @@
 /ql/ @github/codeql-ql-for-ql-reviewers
 
 # Bazel (excluding BUILD.bazel files)
-WORKSPACE.bazel @github/codeql-ci-reviewers
 MODULE.bazel @github/codeql-ci-reviewers
 .bazelversion @github/codeql-ci-reviewers
 .bazelrc @github/codeql-ci-reviewers
@@ -43,3 +43,6 @@ MODULE.bazel @github/codeql-ci-reviewers
 # Misc
 /misc/scripts/accept-expected-changes-from-ci.py @RasmusWL
 /misc/scripts/generate-code-scanning-query-list.py @RasmusWL
+
+# .devcontainer
+/.devcontainer/ @github/codeql-ci-reviewers

Cargo.toml

@@ -7,7 +7,8 @@ members = [
     "ruby/extractor",
     "rust/extractor",
     "rust/extractor/macros",
-    "rust/generate-schema",
+    "rust/ast-generator",
+    "rust/autobuild",
 ]
 
 [patch.crates-io]

MODULE.bazel

@@ -15,60 +15,80 @@ local_path_override(
 # see https://registry.bazel.build/ for a list of available packages
 
 bazel_dep(name = "platforms", version = "0.0.10")
-bazel_dep(name = "rules_go", version = "0.50.0")
+bazel_dep(name = "rules_go", version = "0.50.1")
 bazel_dep(name = "rules_pkg", version = "1.0.1")
 bazel_dep(name = "rules_nodejs", version = "6.2.0-codeql.1")
-bazel_dep(name = "rules_python", version = "0.35.0")
-bazel_dep(name = "bazel_skylib", version = "1.6.1")
-bazel_dep(name = "abseil-cpp", version = "20240116.0", repo_name = "absl")
+bazel_dep(name = "rules_python", version = "0.40.0")
+bazel_dep(name = "rules_shell", version = "0.3.0")
+bazel_dep(name = "bazel_skylib", version = "1.7.1")
+bazel_dep(name = "abseil-cpp", version = "20240116.1", repo_name = "absl")
 bazel_dep(name = "nlohmann_json", version = "3.11.3", repo_name = "json")
 bazel_dep(name = "fmt", version = "10.0.0")
-bazel_dep(name = "rules_kotlin", version = "1.9.4-codeql.1")
-bazel_dep(name = "gazelle", version = "0.38.0")
-bazel_dep(name = "rules_dotnet", version = "0.15.1")
+bazel_dep(name = "rules_kotlin", version = "2.0.0-codeql.1")
+bazel_dep(name = "gazelle", version = "0.40.0")
+bazel_dep(name = "rules_dotnet", version = "0.17.4")
 bazel_dep(name = "googletest", version = "1.14.0.bcr.1")
-bazel_dep(name = "rules_rust", version = "0.50.0")
+bazel_dep(name = "rules_rust", version = "0.52.2")
+bazel_dep(name = "zstd", version = "1.5.5.bcr.1")
 
 bazel_dep(name = "buildifier_prebuilt", version = "6.4.0", dev_dependency = True)
 
-# crate_py but shortened due to Windows file path considerations
-cp = use_extension(
-    "@rules_rust//crate_universe:extension.bzl",
-    "crate",
-    isolate = True,
-)
-cp.from_cargo(
-    name = "py_deps",
-    cargo_lockfile = "//python/extractor/tsg-python:Cargo.lock",
-    manifests = [
-        "//python/extractor/tsg-python:Cargo.toml",
-        "//python/extractor/tsg-python/tsp:Cargo.toml",
-    ],
-)
-use_repo(cp, "py_deps")
+# Keep edition and version approximately in sync with internal repo.
+# the versions there are canonical, the versions here are used for CI in github/codeql, as well as for the vendoring of dependencies.
+RUST_EDITION = "2021"
 
-# deps for ruby+rust, but shortened due to windows file paths
-r = use_extension(
-    "@rules_rust//crate_universe:extension.bzl",
-    "crate",
-    isolate = True,
-)
-r.from_cargo(
-    name = "r",
-    cargo_lockfile = "//:Cargo.lock",
-    manifests = [
-        "//:Cargo.toml",
-        "//ruby/extractor:Cargo.toml",
-        "//rust/extractor:Cargo.toml",
-        "//rust/extractor/macros:Cargo.toml",
-        "//rust/generate-schema:Cargo.toml",
-        "//shared/tree-sitter-extractor:Cargo.toml",
+RUST_VERSION = "1.82.0"
+
+rust = use_extension("@rules_rust//rust:extensions.bzl", "rust")
+rust.toolchain(
+    edition = RUST_EDITION,
+    # We need those extra target triples so that we can build universal binaries on macos
+    extra_target_triples = [
+        "x86_64-apple-darwin",
+        "aarch64-apple-darwin",
     ],
+    versions = [RUST_VERSION],
+)
+use_repo(rust, "rust_toolchains")
+
+register_toolchains("@rust_toolchains//:all")
+
+rust_host_tools = use_extension("@rules_rust//rust:extensions.bzl", "rust_host_tools")
+
+# Don't download a second toolchain as host toolchain, make sure this is the same version as above
+# The host toolchain is used for vendoring dependencies.
+rust_host_tools.host_tools(
+    edition = RUST_EDITION,
+    version = RUST_VERSION,
+)
+
+# deps for python extractor
+# keep in sync by running `misc/bazel/3rdparty/update_cargo_deps.sh`
+py_deps = use_extension("//misc/bazel/3rdparty:py_deps_extension.bzl", "p")
+use_repo(py_deps, "vendor__anyhow-1.0.44", "vendor__cc-1.0.70", "vendor__clap-2.33.3", "vendor__regex-1.5.5", "vendor__smallvec-1.6.1", "vendor__string-interner-0.12.2", "vendor__thiserror-1.0.29", "vendor__tree-sitter-0.20.4", "vendor__tree-sitter-graph-0.7.0")
+
+# deps for ruby+rust
+# keep in sync by running `misc/bazel/3rdparty/update_cargo_deps.sh`
+tree_sitter_extractors_deps = use_extension("//misc/bazel/3rdparty:tree_sitter_extractors_extension.bzl", "r")
+use_repo(tree_sitter_extractors_deps, "vendor__anyhow-1.0.94", "vendor__argfile-0.2.1", "vendor__chrono-0.4.39", "vendor__clap-4.5.23", "vendor__dunce-1.0.5", "vendor__either-1.13.0", "vendor__encoding-0.2.33", "vendor__figment-0.10.19", "vendor__flate2-1.0.35", "vendor__glob-0.3.1", "vendor__globset-0.4.15", "vendor__itertools-0.12.1", "vendor__itertools-0.13.0", "vendor__lazy_static-1.5.0", "vendor__log-0.4.22", "vendor__num-traits-0.2.19", "vendor__num_cpus-1.16.0", "vendor__proc-macro2-1.0.92", "vendor__quote-1.0.37", "vendor__ra_ap_base_db-0.0.248", "vendor__ra_ap_cfg-0.0.248", "vendor__ra_ap_hir-0.0.248", "vendor__ra_ap_hir_def-0.0.248", "vendor__ra_ap_hir_expand-0.0.248", "vendor__ra_ap_ide_db-0.0.248", "vendor__ra_ap_intern-0.0.248", "vendor__ra_ap_load-cargo-0.0.248", "vendor__ra_ap_parser-0.0.248", "vendor__ra_ap_paths-0.0.248", "vendor__ra_ap_project_model-0.0.248", "vendor__ra_ap_span-0.0.248", "vendor__ra_ap_stdx-0.0.248", "vendor__ra_ap_syntax-0.0.248", "vendor__ra_ap_vfs-0.0.248", "vendor__rand-0.8.5", "vendor__rayon-1.10.0", "vendor__regex-1.11.1", "vendor__serde-1.0.216", "vendor__serde_json-1.0.133", "vendor__serde_with-3.11.0", "vendor__stderrlog-0.6.0", "vendor__syn-2.0.90", "vendor__tracing-0.1.41", "vendor__tracing-subscriber-0.3.19", "vendor__tree-sitter-0.24.5", "vendor__tree-sitter-embedded-template-0.23.2", "vendor__tree-sitter-json-0.24.8", "vendor__tree-sitter-ql-0.23.1", "vendor__tree-sitter-ruby-0.23.1", "vendor__triomphe-0.1.14", "vendor__ungrammar-1.16.1")
+
+http_archive = use_repo_rule("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
+
+# rust-analyzer sources needed by the rust ast-generator (see `rust/ast-generator/README.md`)
+http_archive(
+    name = "rust-analyzer-src",
+    build_file = "//rust/ast-generator:BUILD.rust-analyzer-src.bazel",
+    integrity = "sha256-jl4KJmZku+ilMLnuX2NU+qa1v10IauSiDiz23sZo360=",
+    patch_args = ["-p1"],
+    patches = [
+        "//rust/ast-generator:patches/rust-analyzer.patch",
+    ],
+    strip_prefix = "rust-analyzer-2024-12-16",
+    url = "https://github.com/rust-lang/rust-analyzer/archive/refs/tags/2024-12-16.tar.gz",
 )
-use_repo(r, tree_sitter_extractors_deps = "r")
 
 dotnet = use_extension("@rules_dotnet//dotnet:extensions.bzl", "dotnet")
-dotnet.toolchain(dotnet_version = "8.0.101")
+dotnet.toolchain(dotnet_version = "9.0.100")
 use_repo(dotnet, "dotnet_toolchains")
 
 register_toolchains("@dotnet_toolchains//:all")
@@ -91,10 +111,12 @@ use_repo(
     swift_deps,
     "binlog",
     "picosha2",
-    "swift_prebuilt_darwin_x86_64",
-    "swift_prebuilt_linux",
-    "swift_toolchain_linux",
-    "swift_toolchain_macos",
+    "swift-prebuilt-linux",
+    "swift-prebuilt-linux-download-only",
+    "swift-prebuilt-macos",
+    "swift-prebuilt-macos-download-only",
+    "swift-resource-dir-linux",
+    "swift-resource-dir-macos",
 )
 
 node = use_extension("@rules_nodejs//nodejs:extensions.bzl", "node")
@@ -186,16 +208,6 @@ lfs_files(
     executable = True,
 )
 
-lfs_files(
-    name = "swift-resource-dir-linux",
-    srcs = ["//swift/third_party/resource-dir:resource-dir-linux.zip"],
-)
-
-lfs_files(
-    name = "swift-resource-dir-macos",
-    srcs = ["//swift/third_party/resource-dir:resource-dir-macos.zip"],
-)
-
 register_toolchains(
     "@nodejs_toolchains//:all",
 )

WORKSPACE.bazel

@@ -1,2 +0,0 @@
-# please use MODULE.bazel to add dependencies
-# this empty file is required by internal repositories, don't remove it

actions/BUILD.bazel

@@ -0,0 +1,9 @@
+load("//misc/bazel:pkg.bzl", "codeql_pack")
+
+package(default_visibility = ["//visibility:public"])
+
+codeql_pack(
+    name = "actions",
+    srcs = ["//actions/extractor"],
+    experimental = True,
+)

actions/extractor/BUILD.bazel

@@ -0,0 +1,10 @@
+load("//misc/bazel:pkg.bzl", "codeql_pkg_files", "strip_prefix")
+
+codeql_pkg_files(
+    name = "extractor",
+    srcs = [
+        "codeql-extractor.yml",
+    ] + glob(["tools/**"]),
+    strip_prefix = strip_prefix.from_pkg(),
+    visibility = ["//actions:__pkg__"],
+)

actions/extractor/codeql-extractor.yml

@@ -0,0 +1,44 @@
+name: "actions"
+aliases: []
+display_name: "GitHub Actions"
+version: 0.0.1
+column_kind: "utf16"
+unicode_newlines: true
+build_modes:
+  - none
+file_coverage_languages: []
+github_api_languages: []
+scc_languages: []
+file_types:
+  - name: workflow
+    display_name: GitHub Actions workflow files
+    extensions:
+      - .yml
+      - .yaml
+forwarded_extractor_name: javascript
+options:
+  trap:
+    title: TRAP options
+    description: Options about how the extractor handles TRAP files
+    type: object
+    visibility: 3
+    properties:
+      cache:
+        title: TRAP cache options
+        description: Options about how the extractor handles its TRAP cache
+        type: object
+        properties:
+          dir:
+            title: TRAP cache directory
+            description: The directory of the TRAP cache to use
+            type: string
+          bound:
+            title: TRAP cache bound
+            description: A soft limit (in MB) on the size of the TRAP cache
+            type: string
+            pattern: "[0-9]+"
+          write:
+            title: TRAP cache writeable
+            description: Whether to write to the TRAP cache as well as reading it
+            type: string
+            pattern: "(true|TRUE|false|FALSE)"

actions/extractor/tools/autobuild-impl.ps1

@@ -0,0 +1,40 @@
+if (($null -ne $env:LGTM_INDEX_INCLUDE) -or ($null -ne $env:LGTM_INDEX_EXCLUDE) -or ($null -ne $env:LGTM_INDEX_FILTERS)) {
+    Write-Output 'Path filters set. Passing them through to the JavaScript extractor.' 
+} else {
+    Write-Output 'No path filters set. Using the default filters.'
+    $DefaultPathFilters = @(
+        'exclude:**/*',
+        'include:.github/workflows/**/*.yml',
+        'include:.github/workflows/**/*.yaml',
+        'include:**/action.yml',
+        'include:**/action.yaml'
+    )
+
+    $env:LGTM_INDEX_FILTERS = $DefaultPathFilters -join "`n"
+}
+
+# Find the JavaScript extractor directory via `codeql resolve extractor`.
+$CodeQL = Join-Path $env:CODEQL_DIST 'codeql.exe'
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_ROOT = &$CodeQL resolve extractor --language javascript
+if ($LASTEXITCODE -ne 0) {
+    throw 'Failed to resolve JavaScript extractor.'
+}
+
+Write-Output "Found JavaScript extractor at '${env:CODEQL_EXTRACTOR_JAVASCRIPT_ROOT}'."
+
+# Run the JavaScript autobuilder.
+$JavaScriptAutoBuild = Join-Path $env:CODEQL_EXTRACTOR_JAVASCRIPT_ROOT 'tools\autobuild.cmd'
+Write-Output "Running JavaScript autobuilder at '${JavaScriptAutoBuild}'."
+
+# Copy the values of the Actions extractor environment variables to the JavaScript extractor environment variables.
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_DIAGNOSTIC_DIR = $env:CODEQL_EXTRACTOR_ACTIONS_DIAGNOSTIC_DIR
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_LOG_DIR = $env:CODEQL_EXTRACTOR_ACTIONS_LOG_DIR
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_SCRATCH_DIR = $env:CODEQL_EXTRACTOR_ACTIONS_SCRATCH_DIR
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_SOURCE_ARCHIVE_DIR = $env:CODEQL_EXTRACTOR_ACTIONS_SOURCE_ARCHIVE_DIR
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_TRAP_DIR = $env:CODEQL_EXTRACTOR_ACTIONS_TRAP_DIR
+$env:CODEQL_EXTRACTOR_JAVASCRIPT_WIP_DATABASE = $env:CODEQL_EXTRACTOR_ACTIONS_WIP_DATABASE
+
+&$JavaScriptAutoBuild
+if ($LASTEXITCODE -ne 0) {
+    throw "JavaScript autobuilder failed."
+}

actions/extractor/tools/autobuild.cmd

@@ -0,0 +1,3 @@
+@echo off
+rem All of the work is done in the PowerShell script
+powershell.exe %~dp0autobuild-impl.ps1

actions/extractor/tools/autobuild.sh

@@ -0,0 +1,39 @@
+#!/bin/sh
+
+set -eu
+
+DEFAULT_PATH_FILTERS=$(cat << END
+exclude:**/*
+include:.github/workflows/**/*.yml
+include:.github/workflows/**/*.yaml
+include:**/action.yml
+include:**/action.yaml
+END
+)
+
+if [ -n "${LGTM_INDEX_INCLUDE:-}" ] || [ -n "${LGTM_INDEX_EXCLUDE:-}" ] || [ -n "${LGTM_INDEX_FILTERS:-}" ] ; then
+    echo "Path filters set. Passing them through to the JavaScript extractor."
+else
+    echo "No path filters set. Using the default filters."
+    LGTM_INDEX_FILTERS="${DEFAULT_PATH_FILTERS}"
+    export LGTM_INDEX_FILTERS
+fi
+
+# Find the JavaScript extractor directory via `codeql resolve extractor`.
+CODEQL_EXTRACTOR_JAVASCRIPT_ROOT="$($CODEQL_DIST/codeql resolve extractor --language javascript)"
+export CODEQL_EXTRACTOR_JAVASCRIPT_ROOT
+
+echo "Found JavaScript extractor at '${CODEQL_EXTRACTOR_JAVASCRIPT_ROOT}'."
+
+# Run the JavaScript autobuilder
+JAVASCRIPT_AUTO_BUILD="${CODEQL_EXTRACTOR_JAVASCRIPT_ROOT}/tools/autobuild.sh"
+echo "Running JavaScript autobuilder at '${JAVASCRIPT_AUTO_BUILD}'."
+
+# Copy the values of the Actions extractor environment variables to the JavaScript extractor environment variables.
+env CODEQL_EXTRACTOR_JAVASCRIPT_DIAGNOSTIC_DIR="${CODEQL_EXTRACTOR_ACTIONS_DIAGNOSTIC_DIR}" \
+    CODEQL_EXTRACTOR_JAVASCRIPT_LOG_DIR="${CODEQL_EXTRACTOR_ACTIONS_LOG_DIR}" \
+    CODEQL_EXTRACTOR_JAVASCRIPT_SCRATCH_DIR="${CODEQL_EXTRACTOR_ACTIONS_SCRATCH_DIR}" \
+    CODEQL_EXTRACTOR_JAVASCRIPT_SOURCE_ARCHIVE_DIR="${CODEQL_EXTRACTOR_ACTIONS_SOURCE_ARCHIVE_DIR}" \
+    CODEQL_EXTRACTOR_JAVASCRIPT_TRAP_DIR="${CODEQL_EXTRACTOR_ACTIONS_TRAP_DIR}" \
+    CODEQL_EXTRACTOR_JAVASCRIPT_WIP_DATABASE="${CODEQL_EXTRACTOR_ACTIONS_WIP_DATABASE}" \
+    ${JAVASCRIPT_AUTO_BUILD}

actions/ql/lib/CHANGELOG.md

@@ -0,0 +1,5 @@
+## 0.4.0
+
+### New Features
+
+* Initial public preview release

actions/ql/lib/actions.qll

@@ -0,0 +1 @@
+import codeql.actions.Ast

actions/ql/lib/change-notes/released/0.4.0.md

@@ -0,0 +1,5 @@
+## 0.4.0
+
+### New Features
+
+* Initial public preview release

actions/ql/lib/codeql-pack.lock.yml

@@ -0,0 +1,4 @@
+---
+lockVersion: 1.0.0
+dependencies: {}
+compiled: false

actions/ql/lib/codeql-pack.release.yml

@@ -0,0 +1,2 @@
+---
+lastReleaseVersion: 0.4.0

actions/ql/lib/codeql/Locations.qll

@@ -0,0 +1,98 @@
+/** Provides classes for working with locations. */
+
+import files.FileSystem
+import codeql.actions.ast.internal.Ast
+
+bindingset[loc]
+pragma[inline_late]
+private string locationToString(Location loc) {
+  exists(string filepath, int startline, int startcolumn, int endline, int endcolumn |
+    loc.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn) and
+    result = filepath + "@" + startline + ":" + startcolumn + ":" + endline + ":" + endcolumn
+  )
+}
+
+newtype TLocation =
+  TBaseLocation(string filepath, int startline, int startcolumn, int endline, int endcolumn) {
+    exists(File file |
+      file.getAbsolutePath() = filepath and
+      locations_default(_, file, startline, startcolumn, endline, endcolumn)
+    )
+    or
+    exists(ExpressionImpl e |
+      e.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
+    )
+    or
+    filepath = "" and startline = 0 and startcolumn = 0 and endline = 0 and endcolumn = 0
+  }
+
+/**
+ * A location as given by a file, a start line, a start column,
+ * an end line, and an end column.
+ *
+ * For more information about locations see [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
+ */
+class Location extends TLocation, TBaseLocation {
+  string filepath;
+  int startline;
+  int startcolumn;
+  int endline;
+  int endcolumn;
+
+  Location() { this = TBaseLocation(filepath, startline, startcolumn, endline, endcolumn) }
+
+  /** Gets the file for this location. */
+  File getFile() {
+    exists(File file |
+      file.getAbsolutePath() = filepath and
+      result = file
+    )
+  }
+
+  /** Gets the 1-based line number (inclusive) where this location starts. */
+  int getStartLine() { result = startline }
+
+  /** Gets the 1-based column number (inclusive) where this location starts. */
+  int getStartColumn() { result = startcolumn }
+
+  /** Gets the 1-based line number (inclusive) where this.getLocationDefault() location ends. */
+  int getEndLine() { result = endline }
+
+  /** Gets the 1-based column number (inclusive) where this.getLocationDefault() location ends. */
+  int getEndColumn() { result = endcolumn }
+
+  /** Gets the number of lines covered by this location. */
+  int getNumLines() { result = endline - startline + 1 }
+
+  /** Gets a textual representation of this element. */
+  pragma[inline]
+  string toString() { result = locationToString(this) }
+
+  /**
+   * Holds if this element is at the specified location.
+   * The location spans column `startcolumn` of line `startline` to
+   * column `endcolumn` of line `endline` in file `filepath`.
+   * For more information, see
+   * [Providing locations in CodeQL queries](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
+   */
+  predicate hasLocationInfo(string p, int sl, int sc, int el, int ec) {
+    p = filepath and
+    sl = startline and
+    sc = startcolumn and
+    el = endline and
+    ec = endcolumn
+  }
+
+  /** Holds if this location starts strictly before the specified location. */
+  pragma[inline]
+  predicate strictlyBefore(Location other) {
+    this.getStartLine() < other.getStartLine()
+    or
+    this.getStartLine() = other.getStartLine() and this.getStartColumn() < other.getStartColumn()
+  }
+}
+
+/** An entity representing an empty location. */
+class EmptyLocation extends Location {
+  EmptyLocation() { this.hasLocationInfo("", 0, 0, 0, 0) }
+}

actions/ql/lib/codeql/actions/Ast.qll

@@ -0,0 +1,400 @@
+private import codeql.actions.ast.internal.Ast
+private import codeql.Locations
+import codeql.actions.Helper
+
+class AstNode instanceof AstNodeImpl {
+  AstNode getAChildNode() { result = super.getAChildNode() }
+
+  AstNode getParentNode() { result = super.getParentNode() }
+
+  string getAPrimaryQlClass() { result = super.getAPrimaryQlClass() }
+
+  Location getLocation() { result = super.getLocation() }
+
+  string toString() { result = super.toString() }
+
+  Step getEnclosingStep() { result = super.getEnclosingStep() }
+
+  Job getEnclosingJob() { result = super.getEnclosingJob() }
+
+  Event getATriggerEvent() { result = super.getATriggerEvent() }
+
+  Workflow getEnclosingWorkflow() { result = super.getEnclosingWorkflow() }
+
+  CompositeAction getEnclosingCompositeAction() { result = super.getEnclosingCompositeAction() }
+
+  Expression getInScopeEnvVarExpr(string name) { result = super.getInScopeEnvVarExpr(name) }
+
+  ScalarValue getInScopeDefaultValue(string name, string prop) {
+    result = super.getInScopeDefaultValue(name, prop)
+  }
+}
+
+class ScalarValue extends AstNode instanceof ScalarValueImpl {
+  string getValue() { result = super.getValue() }
+}
+
+class Expression extends AstNode instanceof ExpressionImpl {
+  string expression;
+  string rawExpression;
+
+  Expression() {
+    expression = this.getExpression() and
+    rawExpression = this.getRawExpression()
+  }
+
+  string getExpression() { result = expression }
+
+  string getRawExpression() { result = rawExpression }
+
+  string getNormalizedExpression() { result = normalizeExpr(expression) }
+}
+
+/** A common class for `env` in workflow, job or step. */
+abstract class Env extends AstNode instanceof EnvImpl {
+  /** Gets an environment variable value given its name. */
+  ScalarValueImpl getEnvVarValue(string name) { result = super.getEnvVarValue(name) }
+
+  /** Gets an environment variable value. */
+  ScalarValueImpl getAnEnvVarValue() { result = super.getAnEnvVarValue() }
+
+  /** Gets an environment variable expressin given its name. */
+  ExpressionImpl getEnvVarExpr(string name) { result = super.getEnvVarExpr(name) }
+
+  /** Gets an environment variable expression. */
+  ExpressionImpl getAnEnvVarExpr() { result = super.getAnEnvVarExpr() }
+}
+
+/**
+ * A custom composite action. This is a mapping at the top level of an Actions YAML action file.
+ * See https://docs.github.com/en/actions/creating-actions/metadata-syntax-for-github-actions.
+ */
+class CompositeAction extends AstNode instanceof CompositeActionImpl {
+  Runs getRuns() { result = super.getRuns() }
+
+  Outputs getOutputs() { result = super.getOutputs() }
+
+  Expression getAnOutputExpr() { result = super.getAnOutputExpr() }
+
+  Expression getOutputExpr(string outputName) { result = super.getOutputExpr(outputName) }
+
+  Input getAnInput() { result = super.getAnInput() }
+
+  Input getInput(string inputName) { result = super.getInput(inputName) }
+
+  LocalJob getACallerJob() { result = super.getACallerJob() }
+
+  UsesStep getACallerStep() { result = super.getACallerStep() }
+
+  predicate isPrivileged() { super.isPrivileged() }
+}
+
+/**
+ * An Actions workflow. This is a mapping at the top level of an Actions YAML workflow file.
+ * See https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions.
+ */
+class Workflow extends AstNode instanceof WorkflowImpl {
+  Env getEnv() { result = super.getEnv() }
+
+  string getName() { result = super.getName() }
+
+  Job getAJob() { result = super.getAJob() }
+
+  Job getJob(string jobId) { result = super.getJob(jobId) }
+
+  Permissions getPermissions() { result = super.getPermissions() }
+
+  Strategy getStrategy() { result = super.getStrategy() }
+
+  On getOn() { result = super.getOn() }
+}
+
+class ReusableWorkflow extends Workflow instanceof ReusableWorkflowImpl {
+  Outputs getOutputs() { result = super.getOutputs() }
+
+  Expression getAnOutputExpr() { result = super.getAnOutputExpr() }
+
+  Expression getOutputExpr(string outputName) { result = super.getOutputExpr(outputName) }
+
+  Input getAnInput() { result = super.getAnInput() }
+
+  Input getInput(string inputName) { result = super.getInput(inputName) }
+
+  ExternalJob getACaller() { result = super.getACaller() }
+}
+
+class Input extends AstNode instanceof InputImpl { }
+
+class Default extends AstNode instanceof DefaultsImpl {
+  ScalarValue getValue(string name, string prop) { result = super.getValue(name, prop) }
+}
+
+class Outputs extends AstNode instanceof OutputsImpl {
+  Expression getAnOutputExpr() { result = super.getAnOutputExpr() }
+
+  Expression getOutputExpr(string outputName) { result = super.getOutputExpr(outputName) }
+
+  override string toString() { result = "Job outputs node" }
+}
+
+class Permissions extends AstNode instanceof PermissionsImpl {
+  bindingset[perm]
+  string getPermission(string perm) { result = super.getPermission(perm) }
+
+  string getAPermission() { result = super.getAPermission() }
+}
+
+class Strategy extends AstNode instanceof StrategyImpl {
+  Expression getMatrixVarExpr(string varName) { result = super.getMatrixVarExpr(varName) }
+
+  Expression getAMatrixVarExpr() { result = super.getAMatrixVarExpr() }
+}
+
+/**
+ * https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#jobsjob_idneeds
+ */
+class Needs extends AstNode instanceof NeedsImpl {
+  Job getANeededJob() { result = super.getANeededJob() }
+}
+
+class On extends AstNode instanceof OnImpl {
+  Event getAnEvent() { result = super.getAnEvent() }
+}
+
+class Event extends AstNode instanceof EventImpl {
+  string getName() { result = super.getName() }
+
+  string getAnActivityType() { result = super.getAnActivityType() }
+
+  string getAPropertyValue(string prop) { result = super.getAPropertyValue(prop) }
+
+  predicate hasProperty(string prop) { super.hasProperty(prop) }
+
+  predicate isExternallyTriggerable() { super.isExternallyTriggerable() }
+
+  predicate isPrivileged() { super.isPrivileged() }
+}
+
+/**
+ * An Actions job within a workflow.
+ * See https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions#jobs.
+ */
+abstract class Job extends AstNode instanceof JobImpl {
+  string getId() { result = super.getId() }
+
+  Workflow getWorkflow() { result = super.getWorkflow() }
+
+  Job getANeededJob() { result = super.getANeededJob() }
+
+  Outputs getOutputs() { result = super.getOutputs() }
+
+  Expression getAnOutputExpr() { result = super.getAnOutputExpr() }
+
+  Expression getOutputExpr(string outputName) { result = super.getOutputExpr(outputName) }
+
+  Env getEnv() { result = super.getEnv() }
+
+  If getIf() { result = super.getIf() }
+
+  Environment getEnvironment() { result = super.getEnvironment() }
+
+  Permissions getPermissions() { result = super.getPermissions() }
+
+  Strategy getStrategy() { result = super.getStrategy() }
+
+  string getARunsOnLabel() { result = super.getARunsOnLabel() }
+
+  predicate isPrivileged() { super.isPrivileged() }
+
+  predicate isPrivilegedExternallyTriggerable(Event event) {
+    super.isPrivilegedExternallyTriggerable(event)
+  }
+}
+
+abstract class StepsContainer extends AstNode instanceof StepsContainerImpl {
+  Step getAStep() { result = super.getAStep() }
+
+  Step getStep(int i) { result = super.getStep(i) }
+}
+
+/**
+ * An `runs` mapping in a custom composite action YAML.
+ * See https://docs.github.com/en/actions/creating-actions/metadata-syntax-for-github-actions#runs
+ */
+class Runs extends StepsContainer instanceof RunsImpl {
+  CompositeAction getAction() { result = super.getAction() }
+}
+
+/**
+ * An Actions job within a workflow which is composed of steps.
+ * See https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions#jobs.
+ */
+class LocalJob extends Job, StepsContainer instanceof LocalJobImpl { }
+
+/**
+ * A step within an Actions job.
+ * See https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions#jobsjob_idsteps.
+ */
+class Step extends AstNode instanceof StepImpl {
+  string getId() { result = super.getId() }
+
+  Env getEnv() { result = super.getEnv() }
+
+  If getIf() { result = super.getIf() }
+
+  StepsContainer getContainer() { result = super.getContainer() }
+
+  Step getNextStep() { result = super.getNextStep() }
+
+  Step getAFollowingStep() { result = super.getAFollowingStep() }
+}
+
+/**
+ * An If node representing a conditional statement.
+ */
+class If extends AstNode instanceof IfImpl {
+  string getCondition() { result = super.getCondition() }
+
+  Expression getConditionExpr() { result = super.getConditionExpr() }
+
+  string getConditionStyle() { result = super.getConditionStyle() }
+}
+
+/**
+ * An Environemnt node representing a deployment environment.
+ */
+class Environment extends AstNode instanceof EnvironmentImpl {
+  string getName() { result = super.getName() }
+
+  Expression getNameExpr() { result = super.getNameExpr() }
+}
+
+abstract class Uses extends AstNode instanceof UsesImpl {
+  string getCallee() { result = super.getCallee() }
+
+  ScalarValue getCalleeNode() { result = super.getCalleeNode() }
+
+  string getVersion() { result = super.getVersion() }
+
+  int getMajorVersion() { result = super.getMajorVersion() }
+
+  string getArgument(string argName) { result = super.getArgument(argName) }
+
+  Expression getArgumentExpr(string argName) { result = super.getArgumentExpr(argName) }
+}
+
+class UsesStep extends Step, Uses instanceof UsesStepImpl { }
+
+class ExternalJob extends Job, Uses instanceof ExternalJobImpl { }
+
+/**
+ * A `run` field within an Actions job step, which runs command-line programs using an operating system shell.
+ * See https://docs.github.com/en/free-pro-team@latest/actions/reference/workflow-syntax-for-github-actions#jobsjob_idstepsrun.
+ */
+class Run extends Step instanceof RunImpl {
+  ShellScript getScript() { result = super.getScript() }
+
+  Expression getAnScriptExpr() { result = super.getAnScriptExpr() }
+
+  string getWorkingDirectory() { result = super.getWorkingDirectory() }
+
+  string getShell() { result = super.getShell() }
+}
+
+class ShellScript extends ScalarValueImpl instanceof ShellScriptImpl {
+  string getRawScript() { result = super.getRawScript() }
+
+  string getStmt(int i) { result = super.getStmt(i) }
+
+  string getAStmt() { result = super.getAStmt() }
+
+  string getCommand(int i) { result = super.getCommand(i) }
+
+  string getACommand() { result = super.getACommand() }
+
+  string getFileReadCommand(int i) { result = super.getFileReadCommand(i) }
+
+  string getAFileReadCommand() { result = super.getAFileReadCommand() }
+
+  predicate getAssignment(int i, string name, string data) { super.getAssignment(i, name, data) }
+
+  predicate getAnAssignment(string name, string data) { super.getAnAssignment(name, data) }
+
+  predicate getAWriteToGitHubEnv(string name, string data) {
+    super.getAWriteToGitHubEnv(name, data)
+  }
+
+  predicate getAWriteToGitHubOutput(string name, string data) {
+    super.getAWriteToGitHubOutput(name, data)
+  }
+
+  predicate getAWriteToGitHubPath(string data) { super.getAWriteToGitHubPath(data) }
+
+  predicate getAnEnvReachingGitHubOutputWrite(string var, string output_field) {
+    super.getAnEnvReachingGitHubOutputWrite(var, output_field)
+  }
+
+  predicate getACmdReachingGitHubOutputWrite(string cmd, string output_field) {
+    super.getACmdReachingGitHubOutputWrite(cmd, output_field)
+  }
+
+  predicate getAnEnvReachingGitHubEnvWrite(string var, string output_field) {
+    super.getAnEnvReachingGitHubEnvWrite(var, output_field)
+  }
+
+  predicate getACmdReachingGitHubEnvWrite(string cmd, string output_field) {
+    super.getACmdReachingGitHubEnvWrite(cmd, output_field)
+  }
+
+  predicate getAnEnvReachingGitHubPathWrite(string var) {
+    super.getAnEnvReachingGitHubPathWrite(var)
+  }
+
+  predicate getACmdReachingGitHubPathWrite(string cmd) { super.getACmdReachingGitHubPathWrite(cmd) }
+
+  predicate getAnEnvReachingArgumentInjectionSink(string var, string command, string argument) {
+    super.getAnEnvReachingArgumentInjectionSink(var, command, argument)
+  }
+
+  predicate getACmdReachingArgumentInjectionSink(string cmd, string command, string argument) {
+    super.getACmdReachingArgumentInjectionSink(cmd, command, argument)
+  }
+
+  predicate fileToGitHubEnv(string path) { super.fileToGitHubEnv(path) }
+
+  predicate fileToGitHubOutput(string path) { super.fileToGitHubOutput(path) }
+
+  predicate fileToGitHubPath(string path) { super.fileToGitHubPath(path) }
+}
+
+abstract class SimpleReferenceExpression extends AstNode instanceof SimpleReferenceExpressionImpl {
+  string getFieldName() { result = super.getFieldName() }
+
+  AstNode getTarget() { result = super.getTarget() }
+}
+
+class JsonReferenceExpression extends AstNode instanceof JsonReferenceExpressionImpl {
+  string getAccessPath() { result = super.getAccessPath() }
+
+  string getInnerExpression() { result = super.getInnerExpression() }
+}
+
+class GitHubExpression extends SimpleReferenceExpression instanceof GitHubExpressionImpl { }
+
+class SecretsExpression extends SimpleReferenceExpression instanceof SecretsExpressionImpl { }
+
+class StepsExpression extends SimpleReferenceExpression instanceof StepsExpressionImpl {
+  string getStepId() { result = super.getStepId() }
+}
+
+class NeedsExpression extends SimpleReferenceExpression instanceof NeedsExpressionImpl {
+  string getNeededJobId() { result = super.getNeededJobId() }
+}
+
+class JobsExpression extends SimpleReferenceExpression instanceof JobsExpressionImpl { }
+
+class InputsExpression extends SimpleReferenceExpression instanceof InputsExpressionImpl { }
+
+class EnvExpression extends SimpleReferenceExpression instanceof EnvExpressionImpl { }
+
+class MatrixExpression extends SimpleReferenceExpression instanceof MatrixExpressionImpl { }

actions/ql/lib/codeql/actions/Bash.qll

@@ -0,0 +1,722 @@
+private import codeql.actions.Ast
+
+class BashShellScript extends ShellScript {
+  BashShellScript() {
+    exists(Run run |
+      this = run.getScript() and
+      run.getShell().matches(["bash%", "sh"])
+    )
+  }
+
+  private string lineProducer(int i) {
+    result = this.getRawScript().regexpReplaceAll("\\\\\\s*\n", "").splitAt("\n", i)
+  }
+
+  private predicate cmdSubstitutionReplacement(string cmdSubs, string id, int k) {
+    exists(string line | line = this.lineProducer(k) |
+      exists(int i, int j |
+        cmdSubs =
+          // $() cmd substitution
+          line.regexpFind("\\$\\((?:[^()]+|\\((?:[^()]+|\\([^()]*\\))*\\))*\\)", i, j)
+              .regexpReplaceAll("^\\$\\(", "")
+              .regexpReplaceAll("\\)$", "") and
+        id = "cmdsubs:" + k + ":" + i + ":" + j
+      )
+      or
+      exists(int i, int j |
+        // `...` cmd substitution
+        cmdSubs =
+          line.regexpFind("\\`[^\\`]+\\`", i, j)
+              .regexpReplaceAll("^\\`", "")
+              .regexpReplaceAll("\\`$", "") and
+        id = "cmd:" + k + ":" + i + ":" + j
+      )
+    )
+  }
+
+  private predicate rankedCmdSubstitutionReplacements(int i, string old, string new) {
+    old = rank[i](string old2 | this.cmdSubstitutionReplacement(old2, _, _) | old2) and
+    this.cmdSubstitutionReplacement(old, new, _)
+  }
+
+  private predicate doReplaceCmdSubstitutions(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.lineProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doReplaceCmdSubstitutions(line, round - 1, old, middle) and
+      this.rankedCmdSubstitutionReplacements(round, target, replacement) and
+      new = middle.replaceAll(target, replacement)
+    )
+  }
+
+  private string cmdSubstitutedLineProducer(int i) {
+    // script lines where any command substitution has been replaced with a unique placeholder
+    result =
+      max(int round, string new |
+        this.doReplaceCmdSubstitutions(i, round, _, new)
+      |
+        new order by round
+      )
+    or
+    this.cmdSubstitutionReplacement(result, _, i)
+  }
+
+  private predicate quotedStringReplacement(string quotedStr, string id) {
+    exists(string line, int k | line = this.cmdSubstitutedLineProducer(k) |
+      exists(int i, int j |
+        // double quoted string
+        quotedStr = line.regexpFind("\"((?:[^\"\\\\]|\\\\.)*)\"", i, j) and
+        id =
+          "qstr:" + k + ":" + i + ":" + j + ":" + quotedStr.length() + ":" +
+            quotedStr.regexpReplaceAll("[^a-zA-Z0-9]", "")
+      )
+      or
+      exists(int i, int j |
+        // single quoted string
+        quotedStr = line.regexpFind("'((?:\\\\.|[^'\\\\])*)'", i, j) and
+        id =
+          "qstr:" + k + ":" + i + ":" + j + ":" + quotedStr.length() + ":" +
+            quotedStr.regexpReplaceAll("[^a-zA-Z0-9]", "")
+      )
+    )
+  }
+
+  private predicate rankedQuotedStringReplacements(int i, string old, string new) {
+    old = rank[i](string old2 | this.quotedStringReplacement(old2, _) | old2) and
+    this.quotedStringReplacement(old, new)
+  }
+
+  private predicate doReplaceQuotedStrings(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.cmdSubstitutedLineProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doReplaceQuotedStrings(line, round - 1, old, middle) and
+      this.rankedQuotedStringReplacements(round, target, replacement) and
+      new = middle.replaceAll(target, replacement)
+    )
+  }
+
+  private string quotedStringLineProducer(int i) {
+    result =
+      max(int round, string new | this.doReplaceQuotedStrings(i, round, _, new) | new order by round)
+  }
+
+  private string stmtProducer(int i) {
+    result = this.quotedStringLineProducer(i).splitAt(Bash::splitSeparator()).trim() and
+    // when splitting the line with a separator that is not present, the result is the original line which may contain other separators
+    // we only one the split parts that do not contain any of the separators
+    not result.indexOf(Bash::splitSeparator()) > -1
+  }
+
+  private predicate doStmtRestoreQuotedStrings(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.stmtProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doStmtRestoreQuotedStrings(line, round - 1, old, middle) and
+      this.rankedQuotedStringReplacements(round, target, replacement) and
+      new = middle.replaceAll(replacement, target)
+    )
+  }
+
+  private string restoredStmtQuotedStringLineProducer(int i) {
+    result =
+      max(int round, string new |
+        this.doStmtRestoreQuotedStrings(i, round, _, new)
+      |
+        new order by round
+      ) and
+    not result.indexOf("qstr:") > -1
+  }
+
+  private predicate doStmtRestoreCmdSubstitutions(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.restoredStmtQuotedStringLineProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doStmtRestoreCmdSubstitutions(line, round - 1, old, middle) and
+      this.rankedCmdSubstitutionReplacements(round, target, replacement) and
+      new = middle.replaceAll(replacement, target)
+    )
+  }
+
+  override string getStmt(int i) {
+    result =
+      max(int round, string new |
+        this.doStmtRestoreCmdSubstitutions(i, round, _, new)
+      |
+        new order by round
+      ) and
+    not result.indexOf("cmdsubs:") > -1
+  }
+
+  override string getAStmt() { result = this.getStmt(_) }
+
+  private string cmdProducer(int i) {
+    result = this.quotedStringLineProducer(i).splitAt(Bash::separator()).trim() and
+    // when splitting the line with a separator that is not present, the result is the original line which may contain other separators
+    // we only one the split parts that do not contain any of the separators
+    not result.indexOf(Bash::separator()) > -1
+  }
+
+  private predicate doCmdRestoreQuotedStrings(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.cmdProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doCmdRestoreQuotedStrings(line, round - 1, old, middle) and
+      this.rankedQuotedStringReplacements(round, target, replacement) and
+      new = middle.replaceAll(replacement, target)
+    )
+  }
+
+  private string restoredCmdQuotedStringLineProducer(int i) {
+    result =
+      max(int round, string new |
+        this.doCmdRestoreQuotedStrings(i, round, _, new)
+      |
+        new order by round
+      ) and
+    not result.indexOf("qstr:") > -1
+  }
+
+  private predicate doCmdRestoreCmdSubstitutions(int line, int round, string old, string new) {
+    round = 0 and
+    old = this.restoredCmdQuotedStringLineProducer(line) and
+    new = old
+    or
+    round > 0 and
+    exists(string middle, string target, string replacement |
+      this.doCmdRestoreCmdSubstitutions(line, round - 1, old, middle) and
+      this.rankedCmdSubstitutionReplacements(round, target, replacement) and
+      new = middle.replaceAll(replacement, target)
+    )
+  }
+
+  string getCmd(int i) {
+    result =
+      max(int round, string new |
+        this.doCmdRestoreCmdSubstitutions(i, round, _, new)
+      |
+        new order by round
+      ) and
+    not result.indexOf("cmdsubs:") > -1
+  }
+
+  string getACmd() { result = this.getCmd(_) }
+
+  override string getCommand(int i) {
+    // remove redirection
+    result =
+      this.getCmd(i)
+          .regexpReplaceAll("(>|>>|2>|2>>|<|<<<)\\s*[\\{\\}\\$\"'_\\-0-9a-zA-Z]+$", "")
+          .trim() and
+    // exclude variable declarations
+    not result.regexpMatch("^[a-zA-Z0-9\\-_]+=") and
+    // exclude comments
+    not result.trim().indexOf("#") = 0 and
+    // exclude the following keywords
+    not result =
+      [
+        "", "for", "in", "do", "done", "if", "then", "else", "elif", "fi", "while", "until", "case",
+        "esac", "{", "}"
+      ]
+  }
+
+  override string getACommand() { result = this.getCommand(_) }
+
+  override string getFileReadCommand(int i) {
+    result = this.getStmt(i) and
+    result.matches(Bash::fileReadCommand() + "%")
+  }
+
+  override string getAFileReadCommand() { result = this.getFileReadCommand(_) }
+
+  override predicate getAssignment(int i, string name, string data) {
+    exists(string stmt |
+      stmt = this.getStmt(i) and
+      name = stmt.regexpCapture("^([a-zA-Z0-9\\-_]+)=.*", 1) and
+      data = stmt.regexpCapture("^[a-zA-Z0-9\\-_]+=(.*)", 1)
+    )
+  }
+
+  override predicate getAnAssignment(string name, string data) { this.getAssignment(_, name, data) }
+
+  override predicate getAWriteToGitHubEnv(string name, string data) {
+    exists(string raw |
+      Bash::extractFileWrite(this, "GITHUB_ENV", raw) and
+      Bash::extractVariableAndValue(raw, name, data)
+    )
+  }
+
+  override predicate getAWriteToGitHubOutput(string name, string data) {
+    exists(string raw |
+      Bash::extractFileWrite(this, "GITHUB_OUTPUT", raw) and
+      Bash::extractVariableAndValue(raw, name, data)
+    )
+  }
+
+  override predicate getAWriteToGitHubPath(string data) {
+    Bash::extractFileWrite(this, "GITHUB_PATH", data)
+  }
+
+  override predicate getAnEnvReachingGitHubOutputWrite(string var, string output_field) {
+    Bash::envReachingGitHubFileWrite(this, var, "GITHUB_OUTPUT", output_field)
+  }
+
+  override predicate getACmdReachingGitHubOutputWrite(string cmd, string output_field) {
+    Bash::cmdReachingGitHubFileWrite(this, cmd, "GITHUB_OUTPUT", output_field)
+  }
+
+  override predicate getAnEnvReachingGitHubEnvWrite(string var, string output_field) {
+    Bash::envReachingGitHubFileWrite(this, var, "GITHUB_ENV", output_field)
+  }
+
+  override predicate getACmdReachingGitHubEnvWrite(string cmd, string output_field) {
+    Bash::cmdReachingGitHubFileWrite(this, cmd, "GITHUB_ENV", output_field)
+  }
+
+  override predicate getAnEnvReachingGitHubPathWrite(string var) {
+    Bash::envReachingGitHubFileWrite(this, var, "GITHUB_PATH", _)
+  }
+
+  override predicate getACmdReachingGitHubPathWrite(string cmd) {
+    Bash::cmdReachingGitHubFileWrite(this, cmd, "GITHUB_PATH", _)
+  }
+
+  override predicate getAnEnvReachingArgumentInjectionSink(
+    string var, string command, string argument
+  ) {
+    Bash::envReachingArgumentInjectionSink(this, var, command, argument)
+  }
+
+  override predicate getACmdReachingArgumentInjectionSink(
+    string cmd, string command, string argument
+  ) {
+    Bash::cmdReachingArgumentInjectionSink(this, cmd, command, argument)
+  }
+
+  override predicate fileToGitHubEnv(string path) {
+    Bash::fileToFileWrite(this, "GITHUB_ENV", path)
+  }
+
+  override predicate fileToGitHubOutput(string path) {
+    Bash::fileToFileWrite(this, "GITHUB_OUTPUT", path)
+  }
+
+  override predicate fileToGitHubPath(string path) {
+    Bash::fileToFileWrite(this, "GITHUB_PATH", path)
+  }
+}
+
+module Bash {
+  string stmtSeparator() { result = ";" }
+
+  string commandSeparator() { result = ["&&", "||"] }
+
+  string splitSeparator() {
+    result = stmtSeparator() or
+    result = commandSeparator()
+  }
+
+  string redirectionSeparator() { result = [">", ">>", "2>", "2>>", ">&", "2>&", "<", "<<<"] }
+
+  string pipeSeparator() { result = "|" }
+
+  string separator() {
+    result = stmtSeparator() or
+    result = commandSeparator() or
+    result = pipeSeparator()
+  }
+
+  string fileReadCommand() { result = ["<", "cat", "jq", "yq", "tail", "head"] }
+
+  /** Checks if expr is a bash command substitution */
+  bindingset[expr]
+  predicate isCmdSubstitution(string expr, string cmd) {
+    exists(string regexp |
+      // $(cmd)
+      regexp = "\\$\\(([^)]+)\\)" and
+      cmd = expr.regexpCapture(regexp, 1)
+      or
+      // `cmd`
+      regexp = "`([^`]+)`" and
+      cmd = expr.regexpCapture(regexp, 1)
+    )
+  }
+
+  /** Checks if expr is a bash command substitution */
+  bindingset[expr]
+  predicate containsCmdSubstitution(string expr, string cmd) {
+    exists(string regexp |
+      // $(cmd)
+      regexp = ".*\\$\\(([^)]+)\\).*" and
+      cmd = expr.regexpCapture(regexp, 1).trim()
+      or
+      // `cmd`
+      regexp = ".*`([^`]+)`.*" and
+      cmd = expr.regexpCapture(regexp, 1).trim()
+    )
+  }
+
+  /** Checks if expr is a bash parameter expansion */
+  bindingset[expr]
+  predicate isParameterExpansion(string expr, string parameter, string operator, string params) {
+    exists(string regexp |
+      // $VAR
+      regexp = "\\$([a-zA-Z_][a-zA-Z0-9_]+)\\b" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = "" and
+      params = ""
+      or
+      // ${VAR}
+      regexp = "\\$\\{([a-zA-Z_][a-zA-Z0-9_]*)\\}" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = "" and
+      params = ""
+      or
+      // ${!VAR}
+      regexp = "\\$\\{([!#])([a-zA-Z_][a-zA-Z0-9_]*)\\}" and
+      parameter = expr.regexpCapture(regexp, 2) and
+      operator = expr.regexpCapture(regexp, 1) and
+      params = ""
+      or
+      // ${VAR<OP><PARAMS>}, ...
+      regexp = "\\$\\{([a-zA-Z_][a-zA-Z0-9_]*)([#%/:^,\\-+]{1,2})?(.*?)\\}" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = expr.regexpCapture(regexp, 2) and
+      params = expr.regexpCapture(regexp, 3)
+    )
+  }
+
+  bindingset[expr]
+  predicate containsParameterExpansion(string expr, string parameter, string operator, string params) {
+    exists(string regexp |
+      // $VAR
+      regexp = ".*\\$([a-zA-Z_][a-zA-Z0-9_]+)\\b.*" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = "" and
+      params = ""
+      or
+      // ${VAR}
+      regexp = ".*\\$\\{([a-zA-Z_][a-zA-Z0-9_]*)\\}.*" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = "" and
+      params = ""
+      or
+      // ${!VAR}
+      regexp = ".*\\$\\{([!#])([a-zA-Z_][a-zA-Z0-9_]*)\\}.*" and
+      parameter = expr.regexpCapture(regexp, 2) and
+      operator = expr.regexpCapture(regexp, 1) and
+      params = ""
+      or
+      // ${VAR<OP><PARAMS>}, ...
+      regexp = ".*\\$\\{([a-zA-Z_][a-zA-Z0-9_]*)([#%/:^,\\-+]{1,2})?(.*?)\\}.*" and
+      parameter = expr.regexpCapture(regexp, 1) and
+      operator = expr.regexpCapture(regexp, 2) and
+      params = expr.regexpCapture(regexp, 3)
+    )
+  }
+
+  bindingset[raw_content]
+  predicate extractVariableAndValue(string raw_content, string key, string value) {
+    exists(string regexp, string content | content = trimQuotes(raw_content) |
+      regexp = "(?msi).*^([a-zA-Z_][a-zA-Z0-9_]*)\\s*<<\\s*['\"]?(\\S+)['\"]?\\s*\n(.*?)\n\\2\\s*$" and
+      key = trimQuotes(content.regexpCapture(regexp, 1)) and
+      value = trimQuotes(content.regexpCapture(regexp, 3))
+      or
+      exists(string line |
+        line = content.splitAt("\n") and
+        regexp = "(?i)^([a-zA-Z_][a-zA-Z0-9_\\-]*)\\s*=\\s*(.*)$" and
+        key = trimQuotes(line.regexpCapture(regexp, 1)) and
+        value = trimQuotes(line.regexpCapture(regexp, 2))
+      )
+    )
+  }
+
+  bindingset[script]
+  predicate singleLineFileWrite(
+    string script, string cmd, string file, string content, string filters
+  ) {
+    exists(string regexp |
+      regexp = "(?i)(echo|printf)\\s*(.*?)\\s*(>>|>|\\s*\\|\\s*tee\\s*(-a|--append)?)\\s*(\\S+)" and
+      cmd = script.regexpCapture(regexp, 1) and
+      file = trimQuotes(script.regexpCapture(regexp, 5)) and
+      filters = "" and
+      content = script.regexpCapture(regexp, 2)
+    )
+  }
+
+  bindingset[script]
+  predicate singleLineWorkflowCmd(string script, string cmd, string key, string value) {
+    exists(string regexp |
+      regexp = "(?i)(echo|printf)\\s*(['|\"])?::(set-[a-z]+)\\s*name\\s*=\\s*(.*?)::(.*)" and
+      cmd = script.regexpCapture(regexp, 3) and
+      key = script.regexpCapture(regexp, 4) and
+      value = trimQuotes(script.regexpCapture(regexp, 5))
+      or
+      regexp = "(?i)(echo|printf)\\s*(['|\"])?::(add-[a-z]+)\\s*::(.*)" and
+      cmd = script.regexpCapture(regexp, 3) and
+      key = "" and
+      value = trimQuotes(script.regexpCapture(regexp, 4))
+    )
+  }
+
+  bindingset[script]
+  predicate heredocFileWrite(string script, string cmd, string file, string content, string filters) {
+    exists(string regexp |
+      regexp =
+        "(?msi).*^(cat)\\s*(>>|>|\\s*\\|\\s*tee\\s*(-a|--append)?)\\s*(\\S+)\\s*<<\\s*['\"]?(\\S+)['\"]?\\s*\n(.*?)\n\\4\\s*$.*" and
+      cmd = script.regexpCapture(regexp, 1) and
+      file = trimQuotes(script.regexpCapture(regexp, 4)) and
+      content = script.regexpCapture(regexp, 6) and
+      filters = ""
+      or
+      regexp =
+        "(?msi).*^(cat)\\s*(<<|<)\\s*[-]?['\"]?(\\S+)['\"]?\\s*([^>]*)(>>|>|\\s*\\|\\s*tee\\s*(-a|--append)?)\\s*(\\S+)\\s*\n(.*?)\n\\3\\s*$.*" and
+      cmd = script.regexpCapture(regexp, 1) and
+      file = trimQuotes(script.regexpCapture(regexp, 7)) and
+      filters = script.regexpCapture(regexp, 4) and
+      content = script.regexpCapture(regexp, 8)
+    )
+  }
+
+  bindingset[script]
+  predicate linesFileWrite(string script, string cmd, string file, string content, string filters) {
+    exists(string regexp, string var_name |
+      regexp =
+        "(?msi).*((echo|printf)\\s+['|\"]?(.*?<<(\\S+))['|\"]?\\s*>>\\s*(\\S+)\\s*[\r\n]+)" +
+          "(((.*?)\\s*>>\\s*\\S+\\s*[\r\n]+)+)" +
+          "((echo|printf)\\s+['|\"]?(EOF)['|\"]?\\s*>>\\s*\\S+\\s*[\r\n]*).*" and
+      var_name = trimQuotes(script.regexpCapture(regexp, 3)).regexpReplaceAll("<<\\s*(\\S+)", "") and
+      content =
+        var_name + "=$(" +
+          trimQuotes(script.regexpCapture(regexp, 6))
+              .regexpReplaceAll(">>.*GITHUB_(ENV|OUTPUT)(})?", "")
+              .trim() + ")" and
+      cmd = "echo" and
+      file = trimQuotes(script.regexpCapture(regexp, 5)) and
+      filters = ""
+    )
+  }
+
+  bindingset[script]
+  predicate blockFileWrite(string script, string cmd, string file, string content, string filters) {
+    exists(string regexp, string first_line, string var_name |
+      regexp =
+        "(?msi).*^\\s*\\{\\s*[\r\n]" +
+          //
+          "(.*?)" +
+          //
+          "(\\s*\\}\\s*(>>|>|\\s*\\|\\s*tee\\s*(-a|--append)?)\\s*(\\S+))\\s*$.*" and
+      first_line = script.regexpCapture(regexp, 1).splitAt("\n", 0).trim() and
+      var_name = first_line.regexpCapture("echo\\s+('|\\\")?(.*)<<.*", 2) and
+      content = var_name + "=$(" + script.regexpCapture(regexp, 1).splitAt("\n").trim() + ")" and
+      not content.indexOf("EOF") > 0 and
+      file = trimQuotes(script.regexpCapture(regexp, 5)) and
+      cmd = "echo" and
+      filters = ""
+    )
+  }
+
+  bindingset[script]
+  predicate multiLineFileWrite(
+    string script, string cmd, string file, string content, string filters
+  ) {
+    heredocFileWrite(script, cmd, file, content, filters)
+    or
+    linesFileWrite(script, cmd, file, content, filters)
+    or
+    blockFileWrite(script, cmd, file, content, filters)
+  }
+
+  bindingset[file_var]
+  predicate extractFileWrite(BashShellScript script, string file_var, string content) {
+    // single line assignment
+    exists(string file_expr, string raw_content |
+      isParameterExpansion(file_expr, file_var, _, _) and
+      singleLineFileWrite(script.getAStmt(), _, file_expr, raw_content, _) and
+      content = trimQuotes(raw_content)
+    )
+    or
+    // workflow command assignment
+    exists(string key, string value, string cmd |
+      (
+        file_var = "GITHUB_ENV" and
+        cmd = "set-env" and
+        content = key + "=" + value
+        or
+        file_var = "GITHUB_OUTPUT" and
+        cmd = "set-output" and
+        content = key + "=" + value
+        or
+        file_var = "GITHUB_PATH" and
+        cmd = "add-path" and
+        content = value
+      ) and
+      singleLineWorkflowCmd(script.getAStmt(), cmd, key, value)
+    )
+    or
+    // multiline assignment
+    exists(string file_expr, string raw_content |
+      multiLineFileWrite(script.getRawScript(), _, file_expr, raw_content, _) and
+      isParameterExpansion(file_expr, file_var, _, _) and
+      content = trimQuotes(raw_content)
+    )
+  }
+
+  /** Writes the content of the file specified by `path` into a file pointed to by `file_var` */
+  predicate fileToFileWrite(BashShellScript script, string file_var, string path) {
+    exists(string regexp, string stmt, string file_expr |
+      regexp =
+        "(?i)(cat)\\s*" + "((?:(?!<<|<<-)[^>\n])+)\\s*" +
+          "(>>|>|\\s*\\|\\s*tee\\s*(-a|--append)?)\\s*" + "(\\S+)" and
+      stmt = script.getAStmt() and
+      file_expr = trimQuotes(stmt.regexpCapture(regexp, 5)) and
+      path = stmt.regexpCapture(regexp, 2) and
+      containsParameterExpansion(file_expr, file_var, _, _)
+    )
+  }
+
+  /**
+   * Holds if the Run scripts contains an access to an environment variable called `var`
+   * which value may get appended to the GITHUB_XXX special file
+   */
+  predicate envReachingGitHubFileWrite(
+    BashShellScript script, string var, string file_var, string field
+  ) {
+    exists(string file_write_value |
+      (
+        file_var = "GITHUB_ENV" and
+        script.getAWriteToGitHubEnv(field, file_write_value)
+        or
+        file_var = "GITHUB_OUTPUT" and
+        script.getAWriteToGitHubOutput(field, file_write_value)
+        or
+        file_var = "GITHUB_PATH" and
+        field = "PATH" and
+        script.getAWriteToGitHubPath(file_write_value)
+      ) and
+      envReachingRunExpr(script, var, file_write_value)
+    )
+  }
+
+  /**
+   * Holds if and environment variable is used, directly or indirectly, in a Run's step expression.
+   * Where the expression is a string captured from the Run's script.
+   */
+  bindingset[expr]
+  predicate envReachingRunExpr(BashShellScript script, string var, string expr) {
+    exists(string var2, string value2 |
+      // VAR2=${VAR:-default} (var2=value2)
+      // echo "FIELD=${VAR2:-default}" >> $GITHUB_ENV (field, file_write_value)
+      script.getAnAssignment(var2, value2) and
+      containsParameterExpansion(value2, var, _, _) and
+      containsParameterExpansion(expr, var2, _, _)
+    )
+    or
+    // var reaches the file write directly
+    // echo "FIELD=${VAR:-default}" >> $GITHUB_ENV (field, file_write_value)
+    containsParameterExpansion(expr, var, _, _)
+  }
+
+  /**
+   * Holds if the Run scripts contains a command substitution (`cmd`)
+   * which output may get appended to the GITHUB_XXX special file
+   */
+  predicate cmdReachingGitHubFileWrite(
+    BashShellScript script, string cmd, string file_var, string field
+  ) {
+    exists(string file_write_value |
+      (
+        file_var = "GITHUB_ENV" and
+        script.getAWriteToGitHubEnv(field, file_write_value)
+        or
+        file_var = "GITHUB_OUTPUT" and
+        script.getAWriteToGitHubOutput(field, file_write_value)
+        or
+        file_var = "GITHUB_PATH" and
+        field = "PATH" and
+        script.getAWriteToGitHubPath(file_write_value)
+      ) and
+      cmdReachingRunExpr(script, cmd, file_write_value)
+    )
+  }
+
+  predicate envReachingArgumentInjectionSink(
+    BashShellScript script, string source, string command, string argument
+  ) {
+    exists(string cmd, string regex, int command_group, int argument_group |
+      cmd = script.getACommand() and
+      argumentInjectionSinksDataModel(regex, command_group, argument_group) and
+      argument = cmd.regexpCapture(regex, argument_group).trim() and
+      command = cmd.regexpCapture(regex, command_group).trim() and
+      envReachingRunExpr(script, source, argument)
+    )
+  }
+
+  predicate cmdReachingArgumentInjectionSink(
+    BashShellScript script, string source, string command, string argument
+  ) {
+    exists(string cmd, string regex, int command_group, int argument_group |
+      cmd = script.getACommand() and
+      argumentInjectionSinksDataModel(regex, command_group, argument_group) and
+      argument = cmd.regexpCapture(regex, argument_group).trim() and
+      command = cmd.regexpCapture(regex, command_group).trim() and
+      cmdReachingRunExpr(script, source, argument)
+    )
+  }
+
+  /**
+   * Holds if a command output is used, directly or indirectly, in a Run's step expression.
+   * Where the expression is a string captured from the Run's script.
+   */
+  bindingset[expr]
+  predicate cmdReachingRunExpr(BashShellScript script, string cmd, string expr) {
+    // cmd output is assigned to a second variable (var2) and var2 reaches the file write
+    exists(string var2, string value2 |
+      // VAR2=$(cmd)
+      // echo "FIELD=${VAR2:-default}" >> $GITHUB_ENV (field, file_write_value)
+      script.getAnAssignment(var2, value2) and
+      containsCmdSubstitution(value2, cmd) and
+      containsParameterExpansion(expr, var2, _, _) and
+      not varMatchesRegexTest(script, var2, alphaNumericRegex())
+    )
+    or
+    // var reaches the file write directly
+    // echo "FIELD=$(cmd)" >> $GITHUB_ENV (field, file_write_value)
+    containsCmdSubstitution(expr, cmd)
+  }
+
+  /**
+   * Holds if there test command that checks a variable against a regex
+   * eg: `[[ $VAR =~ ^[a-zA-Z0-9_]+$ ]]`
+   */
+  bindingset[var, regex]
+  predicate varMatchesRegexTest(BashShellScript script, string var, string regex) {
+    exists(string lhs, string rhs |
+      lhs = script.getACommand().regexpCapture(".*\\[\\[\\s*(.*?)\\s*=~\\s*(.*?)\\s*\\]\\].*", 1) and
+      containsParameterExpansion(lhs, var, _, _) and
+      rhs = script.getACommand().regexpCapture(".*\\[\\[\\s*(.*?)\\s*=~\\s*(.*?)\\s*\\]\\].*", 2) and
+      trimQuotes(rhs).regexpMatch(regex)
+    )
+  }
+
+  /**
+   * Holds if the given regex is used to match an alphanumeric string
+   * eg: `^[0-9a-zA-Z]{40}$`, `^[0-9]+$` or `^[a-zA-Z0-9_]+$`
+   */
+  string alphaNumericRegex() { result = "^\\^\\[([09azAZ_-]+)\\](\\+|\\{\\d+\\})\\$$" }
+}

actions/ql/lib/codeql/actions/Cfg.qll

@@ -0,0 +1,6 @@
+/** Provides classes representing the control flow graph. */
+
+private import codeql.actions.controlflow.internal.Cfg as CfgInternal
+import CfgInternal::Completion
+import CfgInternal::CfgScope
+import CfgInternal::CfgImpl

actions/ql/lib/codeql/actions/Consistency.ql

@@ -0,0 +1 @@
+import DataFlow::DataFlow::Consistency

actions/ql/lib/codeql/actions/DataFlow.qll

@@ -0,0 +1,22 @@
+/**
+ * Provides classes for performing local (intra-procedural) and
+ * global (inter-procedural) data flow analyses.
+ */
+
+import codeql.Locations
+
+module DataFlow {
+  private import codeql.dataflow.DataFlow
+  private import codeql.actions.dataflow.internal.DataFlowImplSpecific
+  import DataFlowMake<Location, ActionsDataFlow>
+  import codeql.actions.dataflow.internal.DataFlowPublic
+  // debug
+  private import codeql.actions.dataflow.internal.TaintTrackingImplSpecific
+  import codeql.dataflow.internal.DataFlowImplConsistency as DFIC
+
+  module ActionsConsistency implements DFIC::InputSig<Location, ActionsDataFlow> { }
+
+  module Consistency {
+    import DFIC::MakeConsistency<Location, ActionsDataFlow, ActionsTaintTracking, ActionsConsistency>
+  }
+}

actions/ql/lib/codeql/actions/Helper.qll

@@ -0,0 +1,88 @@
+private import codeql.actions.Ast
+private import codeql.Locations
+private import codeql.actions.security.ControlChecks
+import codeql.actions.config.Config
+import codeql.actions.Bash
+import codeql.actions.PowerShell
+
+bindingset[expr]
+string normalizeExpr(string expr) {
+  result =
+    expr.regexpReplaceAll("\\['([a-zA-Z0-9_\\*\\-]+)'\\]", ".$1")
+        .regexpReplaceAll("\\[\"([a-zA-Z0-9_\\*\\-]+)\"\\]", ".$1")
+        .regexpReplaceAll("\\s*\\.\\s*", ".")
+}
+
+bindingset[regex]
+string wrapRegexp(string regex) { result = "\\b" + regex + "\\b" }
+
+bindingset[regex]
+string wrapJsonRegexp(string regex) {
+  result = ["fromJSON\\(\\s*" + regex + "\\s*\\)", "toJSON\\(\\s*" + regex + "\\s*\\)"]
+}
+
+bindingset[str]
+string trimQuotes(string str) {
+  result = str.trim().regexpReplaceAll("^(\"|')", "").regexpReplaceAll("(\"|')$", "")
+}
+
+predicate inPrivilegedContext(AstNode node, Event event) {
+  node.getEnclosingJob().isPrivilegedExternallyTriggerable(event)
+}
+
+predicate inNonPrivilegedContext(AstNode node) {
+  not node.getEnclosingJob().isPrivilegedExternallyTriggerable(_)
+}
+
+string defaultBranchNames() {
+  repositoryDataModel(_, result)
+  or
+  not exists(string default_branch_name | repositoryDataModel(_, default_branch_name)) and
+  result = ["main", "master"]
+}
+
+string getRepoRoot() {
+  exists(Workflow w |
+    w.getLocation().getFile().getRelativePath().indexOf("/.github/workflows") > 0 and
+    result =
+      w.getLocation()
+          .getFile()
+          .getRelativePath()
+          .prefix(w.getLocation().getFile().getRelativePath().indexOf("/.github/workflows") + 1) and
+    // exclude workflow_enum reusable workflows directory root
+    not result.indexOf(".github/workflows/external/") > -1 and
+    not result.indexOf(".github/actions/external/") > -1
+    or
+    not w.getLocation().getFile().getRelativePath().indexOf("/.github/workflows") > 0 and
+    not w.getLocation().getFile().getRelativePath().indexOf(".github/workflows/external/") > -1 and
+    not w.getLocation().getFile().getRelativePath().indexOf(".github/actions/external/") > -1 and
+    result = ""
+  )
+}
+
+bindingset[path]
+string normalizePath(string path) {
+  exists(string trimmed_path | trimmed_path = trimQuotes(path) |
+    // ./foo -> GITHUB_WORKSPACE/foo
+    if path.indexOf("./") = 0
+    then result = path.replaceAll("./", "GITHUB_WORKSPACE/")
+    else
+      // GITHUB_WORKSPACE/foo -> GITHUB_WORKSPACE/foo
+      if path.indexOf("GITHUB_WORKSPACE/") = 0
+      then result = path
+      else
+        // foo -> GITHUB_WORKSPACE/foo
+        if path.regexpMatch("^[^/~].*")
+        then result = "GITHUB_WORKSPACE/" + path.regexpReplaceAll("/$", "")
+        else
+          // ~/foo -> ~/foo
+          // /foo -> /foo
+          result = path
+  )
+}
+
+/**
+ * Holds if the path cache_path is a subpath of the path untrusted_path.
+ */
+bindingset[subpath, path]
+predicate isSubpath(string subpath, string path) { subpath.substring(0, path.length()) = path }

actions/ql/lib/codeql/actions/PowerShell.qll

@@ -0,0 +1,62 @@
+private import codeql.actions.Ast
+
+class PowerShellScript extends ShellScript {
+  PowerShellScript() {
+    exists(Run run |
+      this = run.getScript() and
+      run.getShell().matches("pwsh%")
+    )
+  }
+
+  override string getStmt(int i) { none() }
+
+  override string getAStmt() { none() }
+
+  override string getCommand(int i) { none() }
+
+  override string getACommand() { none() }
+
+  override string getFileReadCommand(int i) { none() }
+
+  override string getAFileReadCommand() { none() }
+
+  override predicate getAssignment(int i, string name, string data) { none() }
+
+  override predicate getAnAssignment(string name, string data) { none() }
+
+  override predicate getAWriteToGitHubEnv(string name, string data) { none() }
+
+  override predicate getAWriteToGitHubOutput(string name, string data) { none() }
+
+  override predicate getAWriteToGitHubPath(string data) { none() }
+
+  override predicate getAnEnvReachingGitHubOutputWrite(string var, string output_field) { none() }
+
+  override predicate getACmdReachingGitHubOutputWrite(string cmd, string output_field) { none() }
+
+  override predicate getAnEnvReachingGitHubEnvWrite(string var, string output_field) { none() }
+
+  override predicate getACmdReachingGitHubEnvWrite(string cmd, string output_field) { none() }
+
+  override predicate getAnEnvReachingGitHubPathWrite(string var) { none() }
+
+  override predicate getACmdReachingGitHubPathWrite(string cmd) { none() }
+
+  override predicate getAnEnvReachingArgumentInjectionSink(
+    string var, string command, string argument
+  ) {
+    none()
+  }
+
+  override predicate getACmdReachingArgumentInjectionSink(
+    string cmd, string command, string argument
+  ) {
+    none()
+  }
+
+  override predicate fileToGitHubEnv(string path) { none() }
+
+  override predicate fileToGitHubOutput(string path) { none() }
+
+  override predicate fileToGitHubPath(string path) { none() }
+}

actions/ql/lib/codeql/actions/TaintTracking.qll

@@ -0,0 +1,13 @@
+/**
+ * Provides classes for performing local (intra-procedural) and
+ * global (inter-procedural) taint-tracking analyses.
+ */
+
+import codeql.Locations
+
+module TaintTracking {
+  private import codeql.actions.dataflow.internal.DataFlowImplSpecific
+  private import codeql.actions.dataflow.internal.TaintTrackingImplSpecific
+  private import codeql.dataflow.TaintTracking
+  import TaintFlowMake<Location, ActionsDataFlow, ActionsTaintTracking>
+}

actions/ql/lib/codeql/actions/Violations Of Best Practices/DefaultableCodeQLInitiatlizeActionQuery.qll

@@ -0,0 +1,32 @@
+private import actions
+
+/**
+ * Holds if workflow step uses the github/codeql-action/init action with no customizations.
+ * e.g.
+ *     - name: Initialize
+ *       uses: github/codeql-action/init@v2
+ *       with:
+ *         languages: ruby, javascript
+ */
+class DefaultableCodeQLInitiatlizeActionQuery extends UsesStep {
+  DefaultableCodeQLInitiatlizeActionQuery() {
+    this.getCallee() = "github/codeql-action/init" and
+    not customizedWorkflowStep(this)
+  }
+}
+
+/**
+ * Holds if the with: part of the workflow step contains any arguments for with: other than "languages".
+ * e.g.
+ *  - name: Initialize CodeQL
+ *       uses: github/codeql-action/init@v3
+ *       with:
+ *         languages: ${{ matrix.language }}
+ *         config-file: ./.github/codeql/${{ matrix.language }}/codeql-config.yml
+ */
+predicate customizedWorkflowStep(UsesStep codeQLInitStep) {
+  exists(string arg |
+    exists(codeQLInitStep.getArgument(arg)) and
+    arg != "languages"
+  )
+}

actions/ql/lib/codeql/actions/ast/internal/Yaml.qll

@@ -0,0 +1,57 @@
+/**
+ * Provides classes for working with YAML data.
+ *
+ * YAML documents are represented as abstract syntax trees whose nodes
+ * are either YAML values or alias nodes referring to another YAML value.
+ */
+
+private import codeql.yaml.Yaml as LibYaml
+
+private module YamlSig implements LibYaml::InputSig {
+  import codeql.Locations
+
+  class LocatableBase extends @yaml_locatable {
+    Location getLocation() {
+      exists(@location_default loc, File f, string p, int sl, int sc, int el, int ec |
+        f.getAbsolutePath() = p and
+        locations_default(loc, f, sl, sc, el, ec) and
+        yaml_locations(this, loc) and
+        result = TBaseLocation(p, sl, sc, el, ec)
+      )
+    }
+
+    string toString() { none() }
+  }
+
+  class NodeBase extends LocatableBase, @yaml_node {
+    NodeBase getChildNode(int i) { yaml(result, _, this, i, _, _) }
+
+    string getTag() { yaml(this, _, _, _, result, _) }
+
+    string getAnchor() { yaml_anchors(this, result) }
+
+    override string toString() { yaml(this, _, _, _, _, result) }
+  }
+
+  class ScalarNodeBase extends NodeBase, @yaml_scalar_node {
+    int getStyle() { yaml_scalars(this, result, _) }
+
+    string getValue() { yaml_scalars(this, _, result) }
+  }
+
+  class CollectionNodeBase extends NodeBase, @yaml_collection_node { }
+
+  class MappingNodeBase extends CollectionNodeBase, @yaml_mapping_node { }
+
+  class SequenceNodeBase extends CollectionNodeBase, @yaml_sequence_node { }
+
+  class AliasNodeBase extends NodeBase, @yaml_alias_node {
+    string getTarget() { yaml_aliases(this, result) }
+  }
+
+  class ParseErrorBase extends LocatableBase, @yaml_error {
+    string getMessage() { yaml_errors(this, result) }
+  }
+}
+
+import LibYaml::Make<YamlSig>

actions/ql/lib/codeql/actions/config/Config.qll

@@ -0,0 +1,147 @@
+import ConfigExtensions as Extensions
+
+/**
+ * MaD models for workflow details
+ * Fields:
+ *    - path: Path to the workflow file
+ *    - trigger: Trigger for the workflow
+ *    - job: Job name
+ *    - secrets_source: Source of secrets
+ *    - permissions: Permissions for the workflow
+ *    - runner: Runner info for the workflow
+ */
+predicate workflowDataModel(
+  string path, string trigger, string job, string secrets_source, string permissions, string runner
+) {
+  Extensions::workflowDataModel(path, trigger, job, secrets_source, permissions, runner)
+}
+
+/**
+ * MaD models for repository details
+ * Fields:
+ *    - visibility: Visibility of the repository
+ *    - default_branch_name: Default branch name
+ */
+predicate repositoryDataModel(string visibility, string default_branch_name) {
+  Extensions::repositoryDataModel(visibility, default_branch_name)
+}
+
+/**
+ * MaD models for context/trigger mapping
+ * Fields:
+ *    - trigger: Trigger for the workflow
+ *    - context_prefix: Prefix for the context
+ */
+predicate contextTriggerDataModel(string trigger, string context_prefix) {
+  Extensions::contextTriggerDataModel(trigger, context_prefix)
+}
+
+/**
+ * MaD models for externally triggerable events
+ * Fields:
+ *    - event: Event name
+ */
+predicate externallyTriggerableEventsDataModel(string event) {
+  Extensions::externallyTriggerableEventsDataModel(event)
+}
+
+private string commandLauncher() { result = ["", "sudo\\s+", "su\\s+", "xvfb-run\\s+"] }
+
+/**
+ * MaD models for poisonable commands
+ * Fields:
+ *    - regexp: Regular expression for matching poisonable commands
+ */
+predicate poisonableCommandsDataModel(string regexp) {
+  exists(string sub_regexp |
+    Extensions::poisonableCommandsDataModel(sub_regexp) and
+    regexp = commandLauncher() + sub_regexp + ".*"
+  )
+}
+
+/**
+ * MaD models for poisonable local scripts
+ * Fields:
+ *    - regexp: Regular expression for matching poisonable local scripts
+ *    - group: Script capture group number for the regular expression
+ */
+predicate poisonableLocalScriptsDataModel(string regexp, int command_group) {
+  exists(string sub_regexp |
+    Extensions::poisonableLocalScriptsDataModel(sub_regexp, command_group) and
+    regexp = commandLauncher() + sub_regexp + ".*"
+  )
+}
+
+/**
+ * MaD models for arguments to commands that execute the given argument.
+ * Fields:
+ *    - regexp: Regular expression for matching argument injections.
+ *    - command_group: capture group for the command.
+ *    - argument_group: capture group for the argument.
+ */
+predicate argumentInjectionSinksDataModel(string regexp, int command_group, int argument_group) {
+  exists(string sub_regexp |
+    Extensions::argumentInjectionSinksDataModel(sub_regexp, command_group, argument_group) and
+    regexp = commandLauncher() + sub_regexp
+  )
+}
+
+/**
+ * MaD models for poisonable actions
+ * Fields:
+ *    - action: action name
+ */
+predicate poisonableActionsDataModel(string action) {
+  Extensions::poisonableActionsDataModel(action)
+}
+
+/**
+ * MaD models for event properties that can be user-controlled.
+ * Fields:
+ *    - property: event property
+ *    - kind: property kind
+ */
+predicate untrustedEventPropertiesDataModel(string property, string kind) {
+  Extensions::untrustedEventPropertiesDataModel(property, kind)
+}
+
+/**
+ * MaD models for vulnerable actions
+ * Fields:
+ *    - action: action name
+ *    - vulnerable_version: vulnerable version
+ *    - vulnerable_sha: vulnerable sha
+ *    - fixed_version: fixed version
+ */
+predicate vulnerableActionsDataModel(
+  string action, string vulnerable_version, string vulnerable_sha, string fixed_version
+) {
+  Extensions::vulnerableActionsDataModel(action, vulnerable_version, vulnerable_sha, fixed_version)
+}
+
+/**
+ * MaD models for immutable actions
+ * Fields:
+ *    - action: action name
+ */
+predicate immutableActionsDataModel(string action) { Extensions::immutableActionsDataModel(action) }
+
+/**
+ * MaD models for untrusted git commands
+ * Fields:
+ *    - cmd_regex: Regular expression for matching untrusted git commands
+ *    - flag: Flag for the command
+ */
+predicate untrustedGitCommandDataModel(string cmd_regex, string flag) {
+  Extensions::untrustedGitCommandDataModel(cmd_regex, flag)
+}
+
+/**
+ * MaD models for untrusted gh commands
+ * Fields:
+ *    - cmd_regex: Regular expression for matching untrusted gh commands
+ *    - flag: Flag for the command
+ */
+predicate untrustedGhCommandDataModel(string cmd_regex, string flag) {
+  Extensions::untrustedGhCommandDataModel(cmd_regex, flag)
+}

actions/ql/lib/codeql/actions/config/ConfigExtensions.qll

@@ -0,0 +1,74 @@
+/**
+ * This module provides extensible predicates for defining MaD models.
+ */
+
+/**
+ * Holds if workflow data model exists for the given parameters.
+ */
+extensible predicate workflowDataModel(
+  string path, string trigger, string job, string secrets_source, string permissions, string runner
+);
+
+/**
+ * Holds if repository data model exists for the given parameters.
+ */
+extensible predicate repositoryDataModel(string visibility, string default_branch_name);
+
+/**
+ * Holds if a context expression starting with context_prefix is available for a given trigger.
+ */
+extensible predicate contextTriggerDataModel(string trigger, string context_prefix);
+
+/**
+ * Holds if a given trigger event can be fired by an external actor.
+ */
+extensible predicate externallyTriggerableEventsDataModel(string event);
+
+/**
+ * Holds for strings that match poisonable commands.
+ */
+extensible predicate poisonableCommandsDataModel(string regexp);
+
+/**
+ * Holds for strings that match poisonable local scripts.
+ */
+extensible predicate poisonableLocalScriptsDataModel(string regexp, int group);
+
+/**
+ * Holds for actions that can be poisoned through local files.
+ */
+extensible predicate poisonableActionsDataModel(string action);
+
+/**
+ * Holds for event properties that can be user-controlled.
+ */
+extensible predicate untrustedEventPropertiesDataModel(string property, string kind);
+
+/**
+ * Holds for arguments to commands that execute the given argument
+ */
+extensible predicate argumentInjectionSinksDataModel(
+  string regexp, int command_group, int argument_group
+);
+
+/**
+ * Holds for actions that are known to be vulnerable.
+ */
+extensible predicate vulnerableActionsDataModel(
+  string action, string vulnerable_version, string vulnerable_sha, string fixed_version
+);
+
+/**
+ * Holds for actions that are known to be immutable.
+ */
+extensible predicate immutableActionsDataModel(string action);
+
+/**
+ * Holds for git commands that may introduce untrusted data when called on an attacker controlled branch.
+ */
+extensible predicate untrustedGitCommandDataModel(string cmd_regex, string flag);
+
+/**
+ * Holds for gh commands that may introduce untrusted data
+ */
+extensible predicate untrustedGhCommandDataModel(string cmd_regex, string flag);

actions/ql/lib/codeql/actions/controlflow/BasicBlocks.qll

@@ -0,0 +1,444 @@
+/** Provides classes representing basic blocks. */
+
+private import codeql.actions.Cfg
+private import codeql.actions.Ast
+private import codeql.Locations
+
+/**
+ * A basic block, that is, a maximal straight-line sequence of control flow nodes
+ * without branches or joins.
+ */
+class BasicBlock extends TBasicBlockStart {
+  /** Gets the scope of this basic block. */
+  final CfgScope getScope() { result = this.getFirstNode().getScope() }
+
+  /** Gets an immediate successor of this basic block, if any. */
+  BasicBlock getASuccessor() { result = this.getASuccessor(_) }
+
+  /** Gets an immediate successor of this basic block of a given type, if any. */
+  BasicBlock getASuccessor(SuccessorType t) {
+    result.getFirstNode() = this.getLastNode().getASuccessor(t)
+  }
+
+  /** Gets an immediate predecessor of this basic block, if any. */
+  BasicBlock getAPredecessor() { result.getASuccessor() = this }
+
+  /** Gets an immediate predecessor of this basic block of a given type, if any. */
+  BasicBlock getAPredecessor(SuccessorType t) { result.getASuccessor(t) = this }
+
+  /** Gets the control flow node at a specific (zero-indexed) position in this basic block. */
+  Node getNode(int pos) { bbIndex(this.getFirstNode(), result, pos) }
+
+  /** Gets a control flow node in this basic block. */
+  Node getANode() { result = this.getNode(_) }
+
+  /** Gets the first control flow node in this basic block. */
+  Node getFirstNode() { this = TBasicBlockStart(result) }
+
+  /** Gets the last control flow node in this basic block. */
+  Node getLastNode() { result = this.getNode(this.length() - 1) }
+
+  /** Gets the length of this basic block. */
+  int length() { result = strictcount(this.getANode()) }
+
+  /**
+   * Holds if this basic block immediately dominates basic block `bb`.
+   *
+   * That is, all paths reaching basic block `bb` from some entry point
+   * basic block must go through this basic block (which is an immediate
+   * predecessor of `bb`).
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     return 0
+   *   end
+   *   return 1
+   * end
+   * ```
+   *
+   * The basic block starting on line 2 immediately dominates the
+   * basic block on line 5 (all paths from the entry point of `m`
+   * to `return 1` must go through the `if` block).
+   */
+  predicate immediatelyDominates(BasicBlock bb) { bbIDominates(this, bb) }
+
+  /**
+   * Holds if this basic block strictly dominates basic block `bb`.
+   *
+   * That is, all paths reaching basic block `bb` from some entry point
+   * basic block must go through this basic block (which must be different
+   * from `bb`).
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     return 0
+   *   end
+   *   return 1
+   * end
+   * ```
+   *
+   * The basic block starting on line 2 strictly dominates the
+   * basic block on line 5 (all paths from the entry point of `m`
+   * to `return 1` must go through the `if` block).
+   */
+  predicate strictlyDominates(BasicBlock bb) { bbIDominates+(this, bb) }
+
+  /**
+   * Holds if this basic block dominates basic block `bb`.
+   *
+   * That is, all paths reaching basic block `bb` from some entry point
+   * basic block must go through this basic block.
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     return 0
+   *   end
+   *   return 1
+   * end
+   * ```
+   *
+   * The basic block starting on line 2 dominates the basic
+   * basic block on line 5 (all paths from the entry point of `m`
+   * to `return 1` must go through the `if` block).
+   */
+  predicate dominates(BasicBlock bb) {
+    bb = this or
+    this.strictlyDominates(bb)
+  }
+
+  /**
+   * Holds if `df` is in the dominance frontier of this basic block.
+   * That is, this basic block dominates a predecessor of `df`, but
+   * does not dominate `df` itself.
+   *
+   * Example:
+   *
+   * ```rb
+   * def m x
+   *   if x < 0
+   *     x = -x
+   *     if x > 10
+   *       x = x - 1
+   *     end
+   *   end
+   *   puts x
+   * end
+   * ```
+   *
+   * The basic block on line 8 is in the dominance frontier
+   * of the basic block starting on line 3 because that block
+   * dominates the basic block on line 4, which is a predecessor of
+   * `puts x`. Also, the basic block starting on line 3 does not
+   * dominate the basic block on line 8.
+   */
+  predicate inDominanceFrontier(BasicBlock df) {
+    this.dominatesPredecessor(df) and
+    not this.strictlyDominates(df)
+  }
+
+  /**
+   * Holds if this basic block dominates a predecessor of `df`.
+   */
+  private predicate dominatesPredecessor(BasicBlock df) { this.dominates(df.getAPredecessor()) }
+
+  /**
+   * Gets the basic block that immediately dominates this basic block, if any.
+   *
+   * That is, all paths reaching this basic block from some entry point
+   * basic block must go through the result, which is an immediate basic block
+   * predecessor of this basic block.
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     return 0
+   *   end
+   *   return 1
+   * end
+   * ```
+   *
+   * The basic block starting on line 2 is an immediate dominator of
+   * the basic block on line 5 (all paths from the entry point of `m`
+   * to `return 1` must go through the `if` block, and the `if` block
+   * is an immediate predecessor of `return 1`).
+   */
+  BasicBlock getImmediateDominator() { bbIDominates(result, this) }
+
+  /**
+   * Holds if this basic block strictly post-dominates basic block `bb`.
+   *
+   * That is, all paths reaching a normal exit point basic block from basic
+   * block `bb` must go through this basic block (which must be different
+   * from `bb`).
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     puts "b"
+   *   end
+   *   puts "m"
+   * end
+   * ```
+   *
+   * The basic block on line 5 strictly post-dominates the basic block on
+   * line 3 (all paths to the exit point of `m` from `puts "b"` must go
+   * through `puts "m"`).
+   */
+  predicate strictlyPostDominates(BasicBlock bb) { bbIPostDominates+(this, bb) }
+
+  /**
+   * Holds if this basic block post-dominates basic block `bb`.
+   *
+   * That is, all paths reaching a normal exit point basic block from basic
+   * block `bb` must go through this basic block.
+   *
+   * Example:
+   *
+   * ```rb
+   * def m b
+   *   if b
+   *     puts "b"
+   *   end
+   *   puts "m"
+   * end
+   * ```
+   *
+   * The basic block on line 5 post-dominates the basic block on line 3
+   * (all paths to the exit point of `m` from `puts "b"` must go through
+   * `puts "m"`).
+   */
+  predicate postDominates(BasicBlock bb) {
+    this.strictlyPostDominates(bb) or
+    this = bb
+  }
+
+  /** Holds if this basic block is in a loop in the control flow graph. */
+  predicate inLoop() { this.getASuccessor+() = this }
+
+  /** Gets a textual representation of this basic block. */
+  string toString() { result = this.getFirstNode().toString() }
+
+  /** Gets the location of this basic block. */
+  Location getLocation() { result = this.getFirstNode().getLocation() }
+}
+
+cached
+private module Cached {
+  /** Internal representation of basic blocks. */
+  cached
+  newtype TBasicBlock = TBasicBlockStart(Node cfn) { startsBB(cfn) }
+
+  /** Holds if `cfn` starts a new basic block. */
+  private predicate startsBB(Node cfn) {
+    not exists(cfn.getAPredecessor()) and exists(cfn.getASuccessor())
+    or
+    cfn.isJoin()
+    or
+    cfn.getAPredecessor().isBranch()
+    or
+    /*
+     * In cases such as
+     *
+     * ```rb
+     * if x or y
+     *     foo
+     * else
+     *     bar
+     * ```
+     *
+     * we have a CFG that looks like
+     *
+     * x --false--> [false] x or y --false--> bar
+     * \                    |
+     *  --true--> y --false--
+     *            \
+     *             --true--> [true] x or y --true--> foo
+     *
+     * and we want to ensure that both `foo` and `bar` start a new basic block,
+     * in order to get a `ConditionalBlock` out of the disjunction.
+     */
+
+    exists(cfn.getAPredecessor(any(BooleanSuccessor s)))
+  }
+
+  /**
+   * Holds if `succ` is a control flow successor of `pred` within
+   * the same basic block.
+   */
+  private predicate intraBBSucc(Node pred, Node succ) {
+    succ = pred.getASuccessor() and
+    not startsBB(succ)
+  }
+
+  /**
+   * Holds if `cfn` is the `i`th node in basic block `bb`.
+   *
+   * In other words, `i` is the shortest distance from a node `bb`
+   * that starts a basic block to `cfn` along the `intraBBSucc` relation.
+   */
+  cached
+  predicate bbIndex(Node bbStart, Node cfn, int i) =
+    shortestDistances(startsBB/1, intraBBSucc/2)(bbStart, cfn, i)
+
+  /**
+   * Holds if the first node of basic block `succ` is a control flow
+   * successor of the last node of basic block `pred`.
+   */
+  private predicate succBB(BasicBlock pred, BasicBlock succ) { succ = pred.getASuccessor() }
+
+  /** Holds if `dom` is an immediate dominator of `bb`. */
+  cached
+  predicate bbIDominates(BasicBlock dom, BasicBlock bb) =
+    idominance(entryBB/1, succBB/2)(_, dom, bb)
+
+  /** Holds if `pred` is a basic block predecessor of `succ`. */
+  private predicate predBB(BasicBlock succ, BasicBlock pred) { succBB(pred, succ) }
+
+  /** Holds if `bb` is an exit basic block that represents normal exit. */
+  private predicate normalExitBB(BasicBlock bb) { bb.getANode().(AnnotatedExitNode).isNormal() }
+
+  /** Holds if `dom` is an immediate post-dominator of `bb`. */
+  cached
+  predicate bbIPostDominates(BasicBlock dom, BasicBlock bb) =
+    idominance(normalExitBB/1, predBB/2)(_, dom, bb)
+
+  /**
+   * Gets the `i`th predecessor of join block `jb`, with respect to some
+   * arbitrary order.
+   */
+  cached
+  JoinBlockPredecessor getJoinBlockPredecessor(JoinBlock jb, int i) {
+    none()
+    /*
+     *    result =
+     *      rank[i + 1](JoinBlockPredecessor jbp |
+     *        jbp = jb.getAPredecessor()
+     *      |
+     *        jbp order by JoinBlockPredecessors::getId(jbp), JoinBlockPredecessors::getSplitString(jbp)
+     *      )
+     */
+
+    }
+
+  cached
+  predicate immediatelyControls(ConditionBlock cb, BasicBlock succ, BooleanSuccessor s) {
+    succ = cb.getASuccessor(s) and
+    forall(BasicBlock pred | pred = succ.getAPredecessor() and pred != cb | succ.dominates(pred))
+  }
+
+  cached
+  predicate controls(ConditionBlock cb, BasicBlock controlled, BooleanSuccessor s) {
+    exists(BasicBlock succ | cb.immediatelyControls(succ, s) | succ.dominates(controlled))
+  }
+}
+
+private import Cached
+
+/** Holds if `bb` is an entry basic block. */
+private predicate entryBB(BasicBlock bb) { bb.getFirstNode() instanceof EntryNode }
+
+/**
+ * An entry basic block, that is, a basic block whose first node is
+ * an entry node.
+ */
+class EntryBasicBlock extends BasicBlock {
+  EntryBasicBlock() { entryBB(this) }
+}
+
+/**
+ * An annotated exit basic block, that is, a basic block whose last node is
+ * an annotated exit node.
+ */
+class AnnotatedExitBasicBlock extends BasicBlock {
+  private boolean normal;
+
+  AnnotatedExitBasicBlock() {
+    exists(AnnotatedExitNode n |
+      n = this.getANode() and
+      if n.isNormal() then normal = true else normal = false
+    )
+  }
+
+  /** Holds if this block represent a normal exit. */
+  final predicate isNormal() { normal = true }
+}
+
+/**
+ * An exit basic block, that is, a basic block whose last node is
+ * an exit node.
+ */
+class ExitBasicBlock extends BasicBlock {
+  ExitBasicBlock() { this.getLastNode() instanceof ExitNode }
+}
+
+/*
+ * private module JoinBlockPredecessors {
+ *  private predicate id(AstNode x, AstNode y) { x = y }
+ *
+ *  private predicate idOf(AstNode x, int y) = equivalenceRelation(id/2)(x, y)
+ *
+ *  int getId(JoinBlockPredecessor jbp) {
+ *    idOf(Ast::toTreeSitter(jbp.getFirstNode().(AstCfgNode).getAstNode()), result)
+ *    or
+ *    idOf(Ast::toTreeSitter(jbp.(EntryBasicBlock).getScope()), result)
+ *  }
+ *
+ *  string getSplitString(JoinBlockPredecessor jbp) {
+ *    result = jbp.getFirstNode().(AstCfgNode).getSplitsString()
+ *    or
+ *    not exists(jbp.getFirstNode().(AstCfgNode).getSplitsString()) and
+ *    result = ""
+ *  }
+ * }
+ */
+
+/** A basic block with more than one predecessor. */
+class JoinBlock extends BasicBlock {
+  JoinBlock() { this.getFirstNode().isJoin() }
+
+  /**
+   * Gets the `i`th predecessor of this join block, with respect to some
+   * arbitrary order.
+   */
+  JoinBlockPredecessor getJoinBlockPredecessor(int i) { result = getJoinBlockPredecessor(this, i) }
+}
+
+/** A basic block that is an immediate predecessor of a join block. */
+class JoinBlockPredecessor extends BasicBlock {
+  JoinBlockPredecessor() { this.getASuccessor() instanceof JoinBlock }
+}
+
+/** A basic block that terminates in a condition, splitting the subsequent control flow. */
+class ConditionBlock extends BasicBlock {
+  ConditionBlock() { this.getLastNode().isCondition() }
+
+  /**
+   * Holds if basic block `succ` is immediately controlled by this basic
+   * block with conditional value `s`. That is, `succ` is an immediate
+   * successor of this block, and `succ` can only be reached from
+   * the callable entry point by going via the `s` edge out of this basic block.
+   */
+  predicate immediatelyControls(BasicBlock succ, BooleanSuccessor s) {
+    immediatelyControls(this, succ, s)
+  }
+
+  /**
+   * Holds if basic block `controlled` is controlled by this basic block with
+   * conditional value `s`. That is, `controlled` can only be reached from
+   * the callable entry point by going via the `s` edge out of this basic block.
+   */
+  predicate controls(BasicBlock controlled, BooleanSuccessor s) { controls(this, controlled, s) }
+}

actions/ql/lib/codeql/actions/controlflow/internal/Cfg.qll

@@ -0,0 +1,316 @@
+private import codeql.actions.Ast
+private import codeql.controlflow.Cfg as CfgShared
+private import codeql.Locations
+
+module Completion {
+  private newtype TCompletion =
+    TSimpleCompletion() or
+    TBooleanCompletion(boolean b) { b in [false, true] } or
+    TReturnCompletion()
+
+  abstract class Completion extends TCompletion {
+    abstract string toString();
+
+    predicate isValidForSpecific(AstNode e) { none() }
+
+    predicate isValidFor(AstNode e) { this.isValidForSpecific(e) }
+
+    abstract SuccessorType getAMatchingSuccessorType();
+  }
+
+  abstract class NormalCompletion extends Completion { }
+
+  class SimpleCompletion extends NormalCompletion, TSimpleCompletion {
+    override string toString() { result = "SimpleCompletion" }
+
+    override predicate isValidFor(AstNode e) { not any(Completion c).isValidForSpecific(e) }
+
+    override NormalSuccessor getAMatchingSuccessorType() { any() }
+  }
+
+  class BooleanCompletion extends NormalCompletion, TBooleanCompletion {
+    boolean value;
+
+    BooleanCompletion() { this = TBooleanCompletion(value) }
+
+    override string toString() { result = "BooleanCompletion(" + value + ")" }
+
+    override predicate isValidForSpecific(AstNode e) { none() }
+
+    override BooleanSuccessor getAMatchingSuccessorType() { result.getValue() = value }
+
+    final boolean getValue() { result = value }
+  }
+
+  class ReturnCompletion extends Completion, TReturnCompletion {
+    override string toString() { result = "ReturnCompletion" }
+
+    override predicate isValidForSpecific(AstNode e) { none() }
+
+    override ReturnSuccessor getAMatchingSuccessorType() { any() }
+  }
+
+  cached
+  private newtype TSuccessorType =
+    TNormalSuccessor() or
+    TBooleanSuccessor(boolean b) { b in [false, true] } or
+    TReturnSuccessor()
+
+  class SuccessorType extends TSuccessorType {
+    string toString() { none() }
+  }
+
+  class NormalSuccessor extends SuccessorType, TNormalSuccessor {
+    override string toString() { result = "successor" }
+  }
+
+  class BooleanSuccessor extends SuccessorType, TBooleanSuccessor {
+    boolean value;
+
+    BooleanSuccessor() { this = TBooleanSuccessor(value) }
+
+    override string toString() { result = value.toString() }
+
+    boolean getValue() { result = value }
+  }
+
+  class ReturnSuccessor extends SuccessorType, TReturnSuccessor {
+    override string toString() { result = "return" }
+  }
+}
+
+module CfgScope {
+  abstract class CfgScope extends AstNode { }
+
+  class WorkflowScope extends CfgScope instanceof Workflow { }
+
+  class CompositeActionScope extends CfgScope instanceof CompositeAction { }
+}
+
+private module Implementation implements CfgShared::InputSig<Location> {
+  import codeql.actions.Ast
+  import Completion
+  import CfgScope
+
+  predicate completionIsNormal(Completion c) { not c instanceof ReturnCompletion }
+
+  // Not using CFG splitting, so the following are just dummy types.
+  private newtype TUnit = Unit()
+
+  additional class SplitKindBase = TUnit;
+
+  additional class Split extends TUnit {
+    abstract string toString();
+  }
+
+  predicate completionIsSimple(Completion c) { c instanceof SimpleCompletion }
+
+  predicate completionIsValidFor(Completion c, AstNode e) { c.isValidFor(e) }
+
+  CfgScope getCfgScope(AstNode e) {
+    exists(AstNode p | p = e.getParentNode() |
+      result = p
+      or
+      not p instanceof CfgScope and result = getCfgScope(p)
+    )
+  }
+
+  additional int maxSplits() { result = 0 }
+
+  predicate scopeFirst(CfgScope scope, AstNode e) {
+    first(scope.(Workflow), e) or
+    first(scope.(CompositeAction), e)
+  }
+
+  predicate scopeLast(CfgScope scope, AstNode e, Completion c) {
+    last(scope.(Workflow), e, c) or
+    last(scope.(CompositeAction), e, c)
+  }
+
+  predicate successorTypeIsSimple(SuccessorType t) { t instanceof NormalSuccessor }
+
+  predicate successorTypeIsCondition(SuccessorType t) { t instanceof BooleanSuccessor }
+
+  SuccessorType getAMatchingSuccessorType(Completion c) { result = c.getAMatchingSuccessorType() }
+
+  predicate isAbnormalExitType(SuccessorType t) { none() }
+}
+
+module CfgImpl = CfgShared::Make<Location, Implementation>;
+
+private import CfgImpl
+private import Completion
+private import CfgScope
+
+private class CompositeActionTree extends StandardPreOrderTree instanceof CompositeAction {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        (
+          child = this.(CompositeAction).getAnInput() or
+          child = this.(CompositeAction).getOutputs() or
+          child = this.(CompositeAction).getRuns()
+        ) and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class RunsTree extends StandardPreOrderTree instanceof Runs {
+  override ControlFlowTree getChildNode(int i) { result = super.getStep(i) }
+}
+
+private class WorkflowTree extends StandardPreOrderTree instanceof Workflow {
+  override ControlFlowTree getChildNode(int i) {
+    if this instanceof ReusableWorkflow
+    then
+      result =
+        rank[i](AstNode child, Location l |
+          (
+            child = this.(ReusableWorkflow).getAnInput() or
+            child = this.(ReusableWorkflow).getOutputs() or
+            child = this.(ReusableWorkflow).getStrategy() or
+            child = this.(ReusableWorkflow).getAJob()
+          ) and
+          l = child.getLocation()
+        |
+          child
+          order by
+            l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+        )
+    else
+      result =
+        rank[i](AstNode child, Location l |
+          (
+            child = super.getStrategy() or
+            child = super.getAJob()
+          ) and
+          l = child.getLocation()
+        |
+          child
+          order by
+            l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+        )
+  }
+}
+
+private class OutputsTree extends StandardPreOrderTree instanceof Outputs {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        child = super.getAnOutputExpr() and l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class StrategyTree extends StandardPreOrderTree instanceof Strategy {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        child = super.getAMatrixVarExpr() and l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class JobTree extends StandardPreOrderTree instanceof LocalJob {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        (
+          child = super.getAStep() or
+          child = super.getOutputs() or
+          child = super.getStrategy()
+        ) and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class ExternalJobTree extends StandardPreOrderTree instanceof ExternalJob {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        (
+          child = super.getArgumentExpr(_) or
+          child = super.getInScopeEnvVarExpr(_) or
+          child = super.getOutputs() or
+          child = super.getStrategy()
+        ) and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class UsesTree extends StandardPreOrderTree instanceof UsesStep {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        (child = super.getArgumentExpr(_) or child = super.getInScopeEnvVarExpr(_)) and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class RunTree extends StandardPreOrderTree instanceof Run {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](AstNode child, Location l |
+        (
+          child = super.getInScopeEnvVarExpr(_) or
+          child = super.getAnScriptExpr() or
+          child = super.getScript()
+        ) and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class ScalarValueTree extends StandardPreOrderTree instanceof ScalarValue {
+  override ControlFlowTree getChildNode(int i) {
+    result =
+      rank[i](Expression child, Location l |
+        child = super.getAChildNode() and
+        l = child.getLocation()
+      |
+        child
+        order by
+          l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(), child.toString()
+      )
+  }
+}
+
+private class UsesLeaf extends LeafTree instanceof Uses { }
+
+private class InputTree extends LeafTree instanceof Input { }
+
+private class ScalarValueLeaf extends LeafTree instanceof ScalarValue { }
+
+private class ExpressionLeaf extends LeafTree instanceof Expression { }

actions/ql/lib/codeql/actions/dataflow/ExternalFlow.qll

@@ -0,0 +1,131 @@
+private import actions
+private import internal.ExternalFlowExtensions as Extensions
+private import codeql.actions.DataFlow
+private import codeql.actions.security.ArtifactPoisoningQuery
+
+/**
+ * MaD sources
+ * Fields:
+ *    - action: Fully-qualified action name (NWO)
+ *    - version: Either '*' or a specific SHA/Tag
+ *    - output arg: To node (prefixed with either `env.` or `output.`)
+ *    - provenance: verification of the model
+ */
+predicate actionsSourceModel(
+  string action, string version, string output, string kind, string provenance
+) {
+  Extensions::actionsSourceModel(action, version, output, kind, provenance)
+}
+
+/**
+ * MaD summaries
+ * Fields:
+ *    - action: Fully-qualified action name (NWO)
+ *    - version: Either '*' or a specific SHA/Tag
+ *    - input arg: From node (prefixed with either `env.` or `input.`)
+ *    - output arg: To node (prefixed with either `env.` or `output.`)
+ *    - kind: Either 'Taint' or 'Value'
+ *    - provenance: verification of the model
+ */
+predicate actionsSummaryModel(
+  string action, string version, string input, string output, string kind, string provenance
+) {
+  Extensions::actionsSummaryModel(action, version, input, output, kind, provenance)
+}
+
+/**
+ * MaD sinks
+ * Fields:
+ *    - action: Fully-qualified action name (NWO)
+ *    - version: Either '*' or a specific SHA/Tag
+ *    - input: sink node (prefixed with either `env.` or `input.`)
+ *    - kind: sink kind
+ *    - provenance: verification of the model
+ */
+predicate actionsSinkModel(
+  string action, string version, string input, string kind, string provenance
+) {
+  Extensions::actionsSinkModel(action, version, input, kind, provenance)
+}
+
+/**
+ * Holds if source.fieldName is a MaD-defined source of a given taint kind.
+ */
+predicate madSource(DataFlow::Node source, string kind, string fieldName) {
+  exists(Uses uses, string action, string version |
+    actionsSourceModel(action, version, fieldName, kind, _) and
+    uses.getCallee() = action.toLowerCase() and
+    (
+      if version.trim() = "*"
+      then uses.getVersion() = any(string v)
+      else uses.getVersion() = version.trim()
+    ) and
+    (
+      if fieldName.trim().matches("env.%")
+      then source.asExpr() = uses.getInScopeEnvVarExpr(fieldName.trim().replaceAll("env.", ""))
+      else
+        if fieldName.trim().matches("output.%")
+        then source.asExpr() = uses
+        else none()
+    )
+  )
+}
+
+/**
+ * Holds if the data flow from `pred` to `succ` is a MaD store step.
+ */
+predicate madStoreStep(DataFlow::Node pred, DataFlow::Node succ, DataFlow::ContentSet c) {
+  exists(Uses uses, string action, string version, string input, string output |
+    actionsSummaryModel(action, version, input, output, "taint", _) and
+    c = any(DataFlow::FieldContent ct | ct.getName() = output.replaceAll("output.", "")) and
+    uses.getCallee() = action.toLowerCase() and
+    // version check
+    (
+      if version.trim() = "*"
+      then uses.getVersion() = any(string v)
+      else uses.getVersion() = version.trim()
+    ) and
+    // pred provenance
+    (
+      input.trim().matches("env.%") and
+      pred.asExpr() = uses.getInScopeEnvVarExpr(input.trim().replaceAll("env.", ""))
+      or
+      input.trim().matches("input.%") and
+      pred.asExpr() = uses.getArgumentExpr(input.trim().replaceAll("input.", ""))
+      or
+      input.trim() = "artifact" and
+      exists(UntrustedArtifactDownloadStep download |
+        pred.asExpr() = download and
+        download.getAFollowingStep() = uses
+      )
+    ) and
+    succ.asExpr() = uses
+  )
+}
+
+/**
+ * Holds if sink is a MaD-defined sink for a given taint kind.
+ */
+predicate madSink(DataFlow::Node sink, string kind) {
+  exists(Uses uses, string action, string version, string input |
+    actionsSinkModel(action, version, input, kind, _) and
+    uses.getCallee() = action.toLowerCase() and
+    // version check
+    (
+      if version.trim() = "*"
+      then uses.getVersion() = any(string v)
+      else uses.getVersion() = version.trim()
+    ) and
+    // pred provenance
+    (
+      input.trim().matches("env.%") and
+      sink.asExpr() = uses.getInScopeEnvVarExpr(input.trim().replaceAll("env.", ""))
+      or
+      input.trim().matches("input.%") and
+      sink.asExpr() = uses.getArgumentExpr(input.trim().replaceAll("input.", ""))
+      or
+      input.trim() = "artifact" and
+      sink.asExpr() = uses
+    )
+  )
+}

actions/ql/lib/codeql/actions/dataflow/FlowSources.qll

@@ -0,0 +1,366 @@
+private import codeql.actions.security.ArtifactPoisoningQuery
+private import codeql.actions.security.UntrustedCheckoutQuery
+private import codeql.actions.config.Config
+private import codeql.actions.dataflow.ExternalFlow
+
+/**
+ * A data flow source.
+ */
+abstract class SourceNode extends DataFlow::Node {
+  /**
+   * Gets a string that represents the source kind with respect to threat modeling.
+   */
+  abstract string getThreatModel();
+}
+
+/** A data flow source of remote user input. */
+abstract class RemoteFlowSource extends SourceNode {
+  /** Gets a string that describes the type of this remote flow source. */
+  abstract string getSourceType();
+
+  /** Gets the event that triggered the source. */
+  abstract string getEventName();
+
+  override string getThreatModel() { result = "remote" }
+}
+
+/**
+ * A data flow source of user input from github context.
+ * eg: github.head_ref
+ */
+class GitHubCtxSource extends RemoteFlowSource {
+  string flag;
+  string event;
+  GitHubExpression e;
+
+  GitHubCtxSource() {
+    this.asExpr() = e and
+    // github.head_ref
+    e.getFieldName() = "head_ref" and
+    flag = "branch" and
+    (
+      event = e.getATriggerEvent().getName() and
+      event = "pull_request_target"
+      or
+      not exists(e.getATriggerEvent()) and
+      event = "unknown"
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  override string getEventName() { result = event }
+}
+
+class GitHubEventCtxSource extends RemoteFlowSource {
+  string flag;
+  string context;
+  string event;
+
+  GitHubEventCtxSource() {
+    exists(Expression e, string regexp |
+      this.asExpr() = e and
+      context = e.getExpression() and
+      (
+        // the context is available for the job trigger events
+        event = e.getATriggerEvent().getName() and
+        exists(string context_prefix |
+          contextTriggerDataModel(event, context_prefix) and
+          normalizeExpr(context).matches("%" + context_prefix + "%")
+        )
+        or
+        not exists(e.getATriggerEvent()) and
+        event = "unknown"
+      ) and
+      untrustedEventPropertiesDataModel(regexp, flag) and
+      not flag = "json" and
+      normalizeExpr(context).regexpMatch("(?i)\\s*" + wrapRegexp(regexp) + ".*")
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  string getContext() { result = context }
+
+  override string getEventName() { result = event }
+}
+
+abstract class CommandSource extends RemoteFlowSource {
+  abstract string getCommand();
+
+  abstract Run getEnclosingRun();
+
+  override string getEventName() { result = this.getEnclosingRun().getATriggerEvent().getName() }
+}
+
+class GitCommandSource extends RemoteFlowSource, CommandSource {
+  Run run;
+  string cmd;
+  string flag;
+
+  GitCommandSource() {
+    exists(Step checkout, string cmd_regex |
+      checkout instanceof SimplePRHeadCheckoutStep and
+      this.asExpr() = run.getScript() and
+      checkout.getAFollowingStep() = run and
+      run.getScript().getAStmt() = cmd and
+      cmd.indexOf("git") = 0 and
+      untrustedGitCommandDataModel(cmd_regex, flag) and
+      cmd.regexpMatch(cmd_regex + ".*")
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  override string getCommand() { result = cmd }
+
+  override Run getEnclosingRun() { result = run }
+}
+
+class GhCLICommandSource extends RemoteFlowSource, CommandSource {
+  Run run;
+  string cmd;
+  string flag;
+
+  GhCLICommandSource() {
+    exists(string cmd_regex |
+      this.asExpr() = run.getScript() and
+      run.getScript().getAStmt() = cmd and
+      cmd.indexOf("gh ") = 0 and
+      untrustedGhCommandDataModel(cmd_regex, flag) and
+      cmd.regexpMatch(cmd_regex + ".*") and
+      (
+        cmd.regexpMatch(".*\\b(pr|pulls)\\b.*") and
+        run.getATriggerEvent().getName() = checkoutTriggers()
+        or
+        not cmd.regexpMatch(".*\\b(pr|pulls)\\b.*")
+      )
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  override Run getEnclosingRun() { result = run }
+
+  override string getCommand() { result = cmd }
+}
+
+class GitHubEventPathSource extends RemoteFlowSource, CommandSource {
+  string cmd;
+  string flag;
+  string access_path;
+  Run run;
+
+  // Examples
+  // COMMENT_AUTHOR=$(jq -r .comment.user.login "$GITHUB_EVENT_PATH")
+  // CURRENT_COMMENT=$(jq -r .comment.body "$GITHUB_EVENT_PATH")
+  // PR_HEAD=$(jq --raw-output .pull_request.head.ref ${GITHUB_EVENT_PATH})
+  // PR_NUMBER=$(jq --raw-output .pull_request.number ${GITHUB_EVENT_PATH})
+  // PR_TITLE=$(jq --raw-output .pull_request.title ${GITHUB_EVENT_PATH})
+  // BODY=$(jq -r '.issue.body' "$GITHUB_EVENT_PATH" | sed -n '3p')
+  GitHubEventPathSource() {
+    this.asExpr() = run.getScript() and
+    run.getScript().getACommand() = cmd and
+    cmd.matches("jq%") and
+    cmd.matches("%GITHUB_EVENT_PATH%") and
+    exists(string regexp |
+      untrustedEventPropertiesDataModel(regexp, flag) and
+      not flag = "json" and
+      access_path = "github.event" + cmd.regexpCapture(".*\\s+([^\\s]+)\\s+.*", 1) and
+      normalizeExpr(access_path).regexpMatch("(?i)\\s*" + wrapRegexp(regexp) + ".*")
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  override string getCommand() { result = cmd }
+
+  override Run getEnclosingRun() { result = run }
+}
+
+class GitHubEventJsonSource extends RemoteFlowSource {
+  string flag;
+  string event;
+
+  GitHubEventJsonSource() {
+    exists(Expression e, string context, string regexp |
+      this.asExpr() = e and
+      context = e.getExpression() and
+      untrustedEventPropertiesDataModel(regexp, _) and
+      (
+        // only contexts for the triggering events are considered tainted.
+        // eg: for `pull_request`, we only consider `github.event.pull_request`
+        event = e.getEnclosingWorkflow().getATriggerEvent().getName() and
+        exists(string context_prefix |
+          contextTriggerDataModel(event, context_prefix) and
+          normalizeExpr(context).matches("%" + context_prefix + "%")
+        ) and
+        normalizeExpr(context).regexpMatch("(?i).*" + wrapJsonRegexp(regexp) + ".*")
+        or
+        // github.event is tainted for all triggers
+        event = e.getEnclosingWorkflow().getATriggerEvent().getName() and
+        contextTriggerDataModel(e.getEnclosingWorkflow().getATriggerEvent().getName(), _) and
+        normalizeExpr(context).regexpMatch("(?i).*" + wrapJsonRegexp("\\bgithub.event\\b") + ".*")
+        or
+        not exists(e.getATriggerEvent()) and
+        event = "unknown"
+      ) and
+      flag = "json"
+    )
+  }
+
+  override string getSourceType() { result = flag }
+
+  override string getEventName() { result = event }
+}
+
+/**
+ * A Source of untrusted data defined in a MaD specification
+ */
+class MaDSource extends RemoteFlowSource {
+  string sourceType;
+
+  MaDSource() { madSource(this, sourceType, _) }
+
+  override string getSourceType() { result = sourceType }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+abstract class FileSource extends RemoteFlowSource { }
+
+/**
+ * A downloaded artifact.
+ */
+class ArtifactSource extends RemoteFlowSource, FileSource {
+  ArtifactSource() { this.asExpr() instanceof UntrustedArtifactDownloadStep }
+
+  override string getSourceType() { result = "artifact" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+/**
+ * A file from an untrusted checkout.
+ */
+private class CheckoutSource extends RemoteFlowSource, FileSource {
+  CheckoutSource() { this.asExpr() instanceof SimplePRHeadCheckoutStep }
+
+  override string getSourceType() { result = "artifact" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+/**
+ * A list of file names returned by dorny/paths-filter.
+ */
+class DornyPathsFilterSource extends RemoteFlowSource {
+  DornyPathsFilterSource() {
+    exists(UsesStep u |
+      u.getCallee() = "dorny/paths-filter" and
+      u.getArgument("list-files") = ["csv", "json"] and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "filename" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+/**
+ * A list of file names returned by tj-actions/changed-files.
+ */
+class TJActionsChangedFilesSource extends RemoteFlowSource {
+  TJActionsChangedFilesSource() {
+    exists(UsesStep u, string vulnerable_action, string vulnerable_version, string vulnerable_sha |
+      vulnerableActionsDataModel(vulnerable_action, vulnerable_version, vulnerable_sha, _) and
+      u.getCallee() = "tj-actions/changed-files" and
+      u.getCallee() = vulnerable_action and
+      (
+        u.getArgument("safe_output") = "false"
+        or
+        (u.getVersion() = vulnerable_version or u.getVersion() = vulnerable_sha)
+      ) and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "filename" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+/**
+ * A list of file names returned by tj-actions/verify-changed-files.
+ */
+class TJActionsVerifyChangedFilesSource extends RemoteFlowSource {
+  TJActionsVerifyChangedFilesSource() {
+    exists(UsesStep u, string vulnerable_action, string vulnerable_version, string vulnerable_sha |
+      vulnerableActionsDataModel(vulnerable_action, vulnerable_version, vulnerable_sha, _) and
+      u.getCallee() = "tj-actions/verify-changed-files" and
+      u.getCallee() = vulnerable_action and
+      (
+        u.getArgument("safe_output") = "false"
+        or
+        (u.getVersion() = vulnerable_version or u.getVersion() = vulnerable_sha)
+      ) and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "filename" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+class Xt0rtedSlashCommandSource extends RemoteFlowSource {
+  Xt0rtedSlashCommandSource() {
+    exists(UsesStep u |
+      u.getCallee() = "xt0rted/slash-command-action" and
+      u.getArgument("permission-level").toLowerCase() = ["read", "none"] and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "text" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+class ZenteredIssueFormBodyParserSource extends RemoteFlowSource {
+  ZenteredIssueFormBodyParserSource() {
+    exists(UsesStep u |
+      u.getCallee() = "zentered/issue-forms-body-parser" and
+      not exists(u.getArgument("body")) and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "text" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}
+
+class OctokitRequestActionSource extends RemoteFlowSource {
+  OctokitRequestActionSource() {
+    exists(UsesStep u, string route |
+      u.getCallee() = "octokit/request-action" and
+      route = u.getArgument("route").trim() and
+      route.indexOf("GET") = 0 and
+      (
+        route.matches("%/commits%") or
+        route.matches("%/comments%") or
+        route.matches("%/pulls%") or
+        route.matches("%/issues%") or
+        route.matches("%/users%") or
+        route.matches("%github.event.issue.pull_request.url%")
+      ) and
+      this.asExpr() = u
+    )
+  }
+
+  override string getSourceType() { result = "text" }
+
+  override string getEventName() { result = this.asExpr().getATriggerEvent().getName() }
+}

actions/ql/lib/codeql/actions/dataflow/FlowSteps.qll

@@ -0,0 +1,92 @@
+/**
+ * Provides classes representing various flow steps for taint tracking.
+ */
+
+private import actions
+private import codeql.actions.DataFlow
+private import codeql.actions.dataflow.FlowSources
+
+/**
+ * Holds if a Run step declares an environment variable, uses it in its script and sets an output in its script.
+ * e.g.
+ *  - name: Extract and Clean Initial URL
+ *    id: extract-url
+ *    env:
+ *      BODY: ${{ github.event.comment.body }}
+ *    run: |
+ *      echo "::set-output name=foo::$BODY"
+ *      echo "foo=$(echo $BODY)" >> $GITHUB_OUTPUT
+ *      echo "foo=$(echo $BODY)" >> "$GITHUB_OUTPUT"
+ *      echo "::set-output name=step-output::$BODY"
+ */
+predicate envToOutputStoreStep(DataFlow::Node pred, DataFlow::Node succ, DataFlow::ContentSet c) {
+  exists(Run run, string var, string field |
+    run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+    succ.asExpr() = run and
+    run.getScript().getAnEnvReachingGitHubOutputWrite(var, field) and
+    c = any(DataFlow::FieldContent ct | ct.getName() = field)
+  )
+}
+
+predicate envToEnvStoreStep(DataFlow::Node pred, DataFlow::Node succ, DataFlow::ContentSet c) {
+  exists(
+    Run run, string var, string field //string key, string value |
+  |
+    run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+    // we store the taint on the enclosing job since the may not exist an implicit env attribute
+    succ.asExpr() = run.getEnclosingJob() and
+    run.getScript().getAnEnvReachingGitHubEnvWrite(var, field) and
+    c = any(DataFlow::FieldContent ct | ct.getName() = field)
+  )
+}
+
+/**
+ * A command whose output gets assigned to an environment variable or step output.
+ * - run: |
+ *     echo "foo=$(cmd)" >> "$GITHUB_OUTPUT"
+ * - run: |
+ *    foo=$(<cmd)"
+ *    echo "bar=${foo}" >> "$GITHUB_OUTPUT"
+ */
+predicate commandToOutputStoreStep(DataFlow::Node pred, DataFlow::Node succ, DataFlow::ContentSet c) {
+  exists(Run run, string key, string cmd |
+    (
+      exists(CommandSource source | source.getCommand() = cmd)
+      or
+      exists(FileSource source |
+        source.asExpr().(Step).getAFollowingStep() = run and
+        run.getScript().getAFileReadCommand() = cmd
+      )
+    ) and
+    run.getScript().getACmdReachingGitHubOutputWrite(cmd, key) and
+    c = any(DataFlow::FieldContent ct | ct.getName() = key) and
+    pred.asExpr() = run.getScript() and
+    succ.asExpr() = run
+  )
+}
+
+/**
+ * A command whose output gets assigned to an environment variable or step output.
+ * - run: |
+ *     echo "foo=$(cmd)" >> "$GITHUB_ENV"
+ * - run: |
+ *    foo=$(<cmd)"
+ *    echo "bar=${foo}" >> "$GITHUB_ENV"
+ */
+predicate commandToEnvStoreStep(DataFlow::Node pred, DataFlow::Node succ, DataFlow::ContentSet c) {
+  exists(Run run, string key, string cmd |
+    (
+      exists(CommandSource source | source.getCommand() = cmd)
+      or
+      exists(FileSource source |
+        source.asExpr().(Step).getAFollowingStep() = run and
+        run.getScript().getAFileReadCommand() = cmd
+      )
+    ) and
+    run.getScript().getACmdReachingGitHubEnvWrite(cmd, key) and
+    c = any(DataFlow::FieldContent ct | ct.getName() = key) and
+    pred.asExpr() = run.getScript() and
+    // we store the taint on the enclosing job since there may not be an implicit env attribute
+    succ.asExpr() = run.getEnclosingJob()
+  )
+}

actions/ql/lib/codeql/actions/dataflow/TaintSteps.qll

@@ -0,0 +1,156 @@
+/**
+ * Provides classes representing various flow steps for taint tracking.
+ */
+
+private import actions
+private import codeql.util.Unit
+private import codeql.actions.DataFlow
+private import codeql.actions.dataflow.FlowSources
+
+/**
+ * A unit class for adding additional taint steps.
+ *
+ * Extend this class to add additional taint steps that should apply to all
+ * taint configurations.
+ */
+class AdditionalTaintStep extends Unit {
+  /**
+   * Holds if the step from `node1` to `node2` should be considered a taint
+   * step for all configurations.
+   */
+  abstract predicate step(DataFlow::Node node1, DataFlow::Node node2);
+}
+
+/**
+ * A file source step followed by a Run step may read the file.
+ */
+predicate fileDownloadToRunStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(FileSource source, Run run |
+    pred = source and
+    source.asExpr().(Step).getAFollowingStep() = run and
+    succ.asExpr() = run.getScript() and
+    exists(run.getScript().getAFileReadCommand())
+  )
+}
+
+/**
+ * A read of the _files field of the dorny/paths-filter action.
+ */
+predicate dornyPathsFilterTaintStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof DornyPathsFilterSource and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    o.getFieldName().matches("%_files") and
+    succ.asExpr() = o
+  )
+}
+
+/**
+ * A read of user-controlled field of the tj-actions/changed-files action.
+ */
+predicate tjActionsChangedFilesTaintStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof TJActionsChangedFilesSource and
+    o.getTarget() = pred.asExpr() and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    o.getFieldName() =
+      [
+        "added_files", "copied_files", "deleted_files", "modified_files", "renamed_files",
+        "all_old_new_renamed_files", "type_changed_files", "unmerged_files", "unknown_files",
+        "all_changed_and_modified_files", "all_changed_files", "other_changed_files",
+        "all_modified_files", "other_modified_files", "other_deleted_files", "modified_keys",
+        "changed_keys"
+      ] and
+    succ.asExpr() = o
+  )
+}
+
+/**
+ * A read of user-controlled field of the tj-actions/verify-changed-files action.
+ */
+predicate tjActionsVerifyChangedFilesTaintStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof TJActionsVerifyChangedFilesSource and
+    o.getTarget() = pred.asExpr() and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    o.getFieldName() = "changed_files" and
+    succ.asExpr() = o
+  )
+}
+
+/**
+ * A read of user-controlled field of the xt0rted/slash-command-action action.
+ */
+predicate xt0rtedSlashCommandActionTaintStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof Xt0rtedSlashCommandSource and
+    o.getTarget() = pred.asExpr() and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    o.getFieldName() = "command-arguments" and
+    succ.asExpr() = o
+  )
+}
+
+/**
+ * A read of user-controlled field of the zentered/issue-forms-body-parser action.
+ */
+predicate zenteredIssueFormBodyParserSource(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof ZenteredIssueFormBodyParserSource and
+    o.getTarget() = pred.asExpr() and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    (
+      not o instanceof JsonReferenceExpression and
+      o.getFieldName() = "data"
+      or
+      o instanceof JsonReferenceExpression and
+      o.(JsonReferenceExpression).getInnerExpression().matches("%.data")
+    ) and
+    succ.asExpr() = o
+  )
+}
+
+/**
+ * A read of user-controlled field of the octokit/request-action action.
+ */
+predicate octokitRequestActionTaintStep(DataFlow::Node pred, DataFlow::Node succ) {
+  exists(StepsExpression o |
+    pred instanceof OctokitRequestActionSource and
+    o.getTarget() = pred.asExpr() and
+    o.getStepId() = pred.asExpr().(UsesStep).getId() and
+    succ.asExpr() = o and
+    (
+      not o instanceof JsonReferenceExpression and
+      o.getFieldName() = "data"
+      or
+      o instanceof JsonReferenceExpression and
+      o.(JsonReferenceExpression).getInnerExpression().matches("%.data") and
+      o.(JsonReferenceExpression)
+          .getAccessPath()
+          .matches([
+              "%.title",
+              "%.user.login",
+              "%.body",
+              "%.head.ref",
+              "%.head.repo.full_name",
+              "%.commit.author.email",
+              "%.commit.commiter.email",
+              "%.commit.message",
+              "%.email",
+              "%.name",
+            ])
+    )
+  )
+}
+
+class TaintSteps extends AdditionalTaintStep {
+  override predicate step(DataFlow::Node node1, DataFlow::Node node2) {
+    dornyPathsFilterTaintStep(node1, node2) or
+    tjActionsChangedFilesTaintStep(node1, node2) or
+    tjActionsVerifyChangedFilesTaintStep(node1, node2) or
+    xt0rtedSlashCommandActionTaintStep(node1, node2) or
+    xt0rtedSlashCommandActionTaintStep(node1, node2) or
+    zenteredIssueFormBodyParserSource(node1, node2) or
+    octokitRequestActionTaintStep(node1, node2)
+  }
+}

actions/ql/lib/codeql/actions/dataflow/internal/DataFlowImplSpecific.qll

@@ -0,0 +1,15 @@
+/**
+ * Provides Actions-specific definitions for use in the data flow library.
+ * Implementation of https://github.com/github/codeql/blob/main/shared/dataflow/codeql/dataflow/DataFlow.qll
+ */
+
+private import codeql.dataflow.DataFlow
+private import codeql.Locations
+
+module ActionsDataFlow implements InputSig<Location> {
+  import DataFlowPrivate as Private
+  import DataFlowPublic
+  import Private
+
+  predicate neverSkipInPathGraph = Private::neverSkipInPathGraph/1;
+}

actions/ql/lib/codeql/actions/dataflow/internal/DataFlowPrivate.qll

@@ -0,0 +1,403 @@
+private import codeql.util.Unit
+private import codeql.dataflow.DataFlow
+private import codeql.actions.Ast
+private import codeql.actions.Cfg as Cfg
+private import codeql.Locations
+private import codeql.actions.controlflow.BasicBlocks
+private import DataFlowPublic
+private import codeql.actions.dataflow.ExternalFlow
+private import codeql.actions.dataflow.FlowSteps
+private import codeql.actions.dataflow.FlowSources
+
+class DataFlowSecondLevelScope = Unit;
+
+cached
+newtype TNode = TExprNode(DataFlowExpr e)
+
+class OutNode extends ExprNode {
+  private DataFlowCall call;
+
+  OutNode() { call = this.getCfgNode() }
+
+  DataFlowCall getCall(ReturnKind kind) {
+    result = call and
+    kind instanceof NormalReturn
+  }
+}
+
+/**
+ * Not implemented
+ */
+class CastNode extends Node {
+  CastNode() { none() }
+}
+
+/**
+ * Not implemented
+ */
+class PostUpdateNode extends Node {
+  PostUpdateNode() { none() }
+
+  Node getPreUpdateNode() { none() }
+}
+
+predicate isParameterNode(ParameterNode p, DataFlowCallable c, ParameterPosition pos) {
+  p.isParameterOf(c, pos)
+}
+
+predicate isArgumentNode(ArgumentNode arg, DataFlowCall call, ArgumentPosition pos) {
+  arg.argumentOf(call, pos)
+}
+
+DataFlowCallable nodeGetEnclosingCallable(Node node) {
+  node = TExprNode(any(DataFlowExpr e | result = e.getScope()))
+}
+
+DataFlowType getNodeType(Node node) { any() }
+
+predicate nodeIsHidden(Node node) { none() }
+
+class DataFlowExpr extends Cfg::Node {
+  DataFlowExpr() {
+    this.getAstNode() instanceof Job or
+    this.getAstNode() instanceof Expression or
+    this.getAstNode() instanceof Uses or
+    this.getAstNode() instanceof Run or
+    this.getAstNode() instanceof Outputs or
+    this.getAstNode() instanceof Input or
+    this.getAstNode() instanceof ScalarValue
+  }
+}
+
+/**
+ * A call corresponds to a Uses steps where a composite action or a reusable workflow get called
+ */
+class DataFlowCall instanceof Cfg::Node {
+  DataFlowCall() { super.getAstNode() instanceof Uses }
+
+  /** Gets a textual representation of this element. */
+  string toString() { result = super.toString() }
+
+  string getName() { result = super.getAstNode().(Uses).getCallee() }
+
+  DataFlowCallable getEnclosingCallable() { result = super.getScope() }
+
+  /** Gets a best-effort total ordering. */
+  int totalorder() { none() }
+
+  /** Gets the location of this call. */
+  Location getLocation() { result = this.(Cfg::Node).getLocation() }
+}
+
+/**
+ * A Cfg scope that can be called
+ */
+class DataFlowCallable instanceof Cfg::CfgScope {
+  string toString() { result = super.toString() }
+
+  string getName() {
+    result = this.(ReusableWorkflowImpl).getResolvedPath() or
+    result = this.(CompositeActionImpl).getResolvedPath()
+  }
+
+  /** Gets a best-effort total ordering. */
+  int totalorder() { none() }
+
+  /** Gets the location of this callable. */
+  Location getLocation() { result = this.(Cfg::CfgScope).getLocation() }
+}
+
+newtype TReturnKind = TNormalReturn()
+
+abstract class ReturnKind extends TReturnKind {
+  /** Gets a textual representation of this element. */
+  abstract string toString();
+}
+
+class NormalReturn extends ReturnKind, TNormalReturn {
+  override string toString() { result = "return" }
+}
+
+/** Gets a viable implementation of the target of the given `Call`. */
+DataFlowCallable viableCallable(DataFlowCall c) { c.getName() = result.getName() }
+
+/**
+ * Gets a node that can read the value returned from `call` with return kind
+ * `kind`.
+ */
+OutNode getAnOutNode(DataFlowCall call, ReturnKind kind) { call = result.getCall(kind) }
+
+private newtype TDataFlowType = TUnknownDataFlowType()
+
+/**
+ * A type for a data flow node.
+ *
+ * This may or may not coincide with any type system existing for the source
+ * language, but should minimally include unique types for individual closure
+ * expressions (typically lambdas).
+ */
+class DataFlowType extends TDataFlowType {
+  string toString() { result = "" }
+}
+
+string ppReprType(DataFlowType t) { none() }
+
+predicate compatibleTypes(DataFlowType t1, DataFlowType t2) { any() }
+
+predicate typeStrongerThan(DataFlowType t1, DataFlowType t2) { none() }
+
+newtype TContent =
+  TFieldContent(string name) {
+    // We only use field flow for env, steps and jobs outputs
+    // not for accessing other context fields such as matrix or inputs
+    name = any(StepsExpression a).getFieldName() or
+    name = any(NeedsExpression a).getFieldName() or
+    name = any(JobsExpression a).getFieldName() or
+    name = any(EnvExpression a).getFieldName()
+  }
+
+predicate forceHighPrecision(Content c) { c instanceof FieldContent }
+
+class NodeRegion instanceof Unit {
+  string toString() { result = "NodeRegion" }
+
+  predicate contains(Node n) { none() }
+
+  int totalOrder() { result = 1 }
+}
+
+/**
+ * Holds if the nodes in `nr` are unreachable when the call context is `call`.
+ */
+predicate isUnreachableInCall(NodeRegion nr, DataFlowCall call) { none() }
+
+class ContentApprox = ContentSet;
+
+ContentApprox getContentApprox(Content c) { result = c }
+
+/**
+ * Made a string to match the ArgumentPosition type.
+ */
+class ParameterPosition extends string {
+  ParameterPosition() {
+    exists(any(ReusableWorkflow w).getInput(this)) or
+    exists(any(CompositeAction a).getInput(this))
+  }
+}
+
+/**
+ * Made a string to match `With:` keys in the AST
+ */
+class ArgumentPosition extends string {
+  ArgumentPosition() { exists(any(Uses e).getArgumentExpr(this)) }
+}
+
+/**
+ */
+predicate parameterMatch(ParameterPosition ppos, ArgumentPosition apos) { ppos = apos }
+
+/**
+ * Holds if there is a local flow step between a ${{ steps.xxx.outputs.yyy }} expression accesing a step output field
+ * and the step output itself. But only for those cases where the step output is defined externally in a MaD Source
+ * specification. The reason for this is that we don't currently have a way to specify that a source starts with a
+ * non-empty access path so we cannot write a Source that stores the taint in a Content, we can only do that for steps
+ * (storeStep). The easiest thing is to add this local flow step that simulates a read step from the source node for a specific
+ * field name.
+ */
+predicate stepsCtxLocalStep(Node nodeFrom, Node nodeTo) {
+  exists(Uses astFrom, StepsExpression astTo |
+    madSource(nodeFrom, _, "output." + ["*", astTo.getFieldName()]) and
+    astFrom = nodeFrom.asExpr() and
+    astTo = nodeTo.asExpr() and
+    astTo.getTarget() = astFrom
+  )
+}
+
+/**
+ * Holds if there is a local flow step between a ${{ needs.xxx.outputs.yyy }} expression accesing a job output field
+ * and the step output itself. But only for those cases where the job (needs) output is defined externally in a MaD Source
+ * specification. The reason for this is that we don't currently have a way to specify that a source starts with a
+ * non-empty access path so we cannot write a Source that stores the taint in a Content, we can only do that for steps
+ * (storeStep). The easiest thing is to add this local flow step that simulates a read step from the source node for a specific
+ * field name.
+ */
+predicate needsCtxLocalStep(Node nodeFrom, Node nodeTo) {
+  exists(Uses astFrom, NeedsExpression astTo |
+    madSource(nodeFrom, _, "output." + astTo.getFieldName()) and
+    astFrom = nodeFrom.asExpr() and
+    astTo = nodeTo.asExpr() and
+    astTo.getTarget() = astFrom
+  )
+}
+
+/**
+ * Holds if there is a local flow step between a ${{}} expression accesing an input variable and the input itself
+ * e.g. ${{ inputs.foo }}
+ */
+predicate inputsCtxLocalStep(Node nodeFrom, Node nodeTo) {
+  exists(AstNode astFrom, InputsExpression astTo |
+    astFrom = nodeFrom.asExpr() and
+    astTo = nodeTo.asExpr() and
+    astTo.getTarget() = astFrom
+  )
+}
+
+/**
+ * Holds if there is a local flow step between a ${{}} expression accesing a matrix variable and the matrix itself
+ * e.g. ${{ matrix.foo }}
+ */
+predicate matrixCtxLocalStep(Node nodeFrom, Node nodeTo) {
+  exists(AstNode astFrom, MatrixExpression astTo |
+    astFrom = nodeFrom.asExpr() and
+    astTo = nodeTo.asExpr() and
+    astTo.getTarget() = astFrom
+  )
+}
+
+/**
+ * Holds if there is a local flow step between a ${{}} expression accesing an env var and the var definition itself
+ * e.g. ${{ env.foo }}
+ */
+predicate envCtxLocalStep(Node nodeFrom, Node nodeTo) {
+  exists(AstNode astFrom, EnvExpression astTo |
+    astFrom = nodeFrom.asExpr() and
+    astTo = nodeTo.asExpr() and
+    (
+      madSource(nodeFrom, _, "env." + astTo.getFieldName())
+      or
+      astTo.getTarget() = astFrom
+    )
+  )
+}
+
+/**
+ * Holds if there is a local flow step from `nodeFrom` to `nodeTo`.
+ * For Actions, we dont need SSA nodes since it should be already in SSA form
+ * Local flow steps are always between two nodes in the same Cfg scope.
+ */
+pragma[nomagic]
+predicate localFlowStep(Node nodeFrom, Node nodeTo) {
+  stepsCtxLocalStep(nodeFrom, nodeTo) or
+  needsCtxLocalStep(nodeFrom, nodeTo) or
+  inputsCtxLocalStep(nodeFrom, nodeTo) or
+  matrixCtxLocalStep(nodeFrom, nodeTo) or
+  envCtxLocalStep(nodeFrom, nodeTo)
+}
+
+/**
+ * This is the local flow predicate that is used as a building block in global
+ * data flow.
+ */
+cached
+predicate simpleLocalFlowStep(Node nodeFrom, Node nodeTo, string model) {
+  localFlowStep(nodeFrom, nodeTo) and model = ""
+}
+
+/**
+ * Holds if data can flow from `node1` to `node2` through a non-local step
+ * that does not follow a call edge. For example, a step through a global
+ * variable.
+ * We throw away the call context and let us jump to any location
+ * AKA teleport steps
+ * local steps are preferible since they are more predictable and easier to control
+ */
+predicate jumpStep(Node nodeFrom, Node nodeTo) { none() }
+
+/**
+ * Holds if a Expression reads a field from a job (needs/jobs), step (steps) output via a read of `c` (fieldname)
+ */
+predicate ctxFieldReadStep(Node node1, Node node2, ContentSet c) {
+  exists(SimpleReferenceExpression access |
+    (
+      access instanceof NeedsExpression or
+      access instanceof StepsExpression or
+      access instanceof JobsExpression or
+      access instanceof EnvExpression
+    ) and
+    c = any(FieldContent ct | ct.getName() = access.getFieldName()) and
+    node1.asExpr() = access.getTarget() and
+    node2.asExpr() = access
+  )
+}
+
+/**
+ * Holds if data can flow from `node1` to `node2` via a read of `c`.  Thus,
+ * `node1` references an object with a content `c.getAReadContent()` whose
+ * value ends up in `node2`.
+ * Store steps without corresponding reads are pruned aggressively very early, since they can never contribute to a complete path.
+ */
+predicate readStep(Node node1, ContentSet c, Node node2) { ctxFieldReadStep(node1, node2, c) }
+
+/**
+ * Stores an output expression (node1) into its OutputsStm node (node2)
+ * using the output variable name as the access path
+ */
+predicate fieldStoreStep(Node node1, Node node2, ContentSet c) {
+  exists(Outputs out, string fieldName |
+    node1.asExpr() = out.getOutputExpr(fieldName) and
+    node2.asExpr() = out and
+    c = any(FieldContent ct | ct.getName() = fieldName)
+  )
+}
+
+/**
+ * Holds if data can flow from `node1` to `node2` via a store into `c`.  Thus,
+ * `node2` references an object with a content `c.getAStoreContent()` that
+ * contains the value of `node1`.
+ * Store steps without corresponding reads are pruned aggressively very early, since they can never contribute to a complete path.
+ */
+predicate storeStep(Node node1, ContentSet c, Node node2) {
+  fieldStoreStep(node1, node2, c) or
+  madStoreStep(node1, node2, c) or
+  envToOutputStoreStep(node1, node2, c) or
+  envToEnvStoreStep(node1, node2, c) or
+  commandToOutputStoreStep(node1, node2, c) or
+  commandToEnvStoreStep(node1, node2, c)
+}
+
+/**
+ * Holds if values stored inside content `c` are cleared at node `n`. For example,
+ * any value stored inside `f` is cleared at the pre-update node associated with `x`
+ * in `x.f = newValue`.
+ */
+predicate clearsContent(Node n, ContentSet c) { none() }
+
+/**
+ * Holds if the value that is being tracked is expected to be stored inside content `c`
+ * at node `n`.
+ */
+predicate expectsContent(Node n, ContentSet c) { none() }
+
+/**
+ * Holds if flow is allowed to pass from parameter `p` and back to itself as a
+ * side-effect, resulting in a summary from `p` to itself.
+ *
+ * One example would be to allow flow like `p.foo = p.bar;`, which is disallowed
+ * by default as a heuristic.
+ */
+predicate allowParameterReturnInSelf(ParameterNode p) { none() }
+
+predicate localMustFlowStep(Node nodeFrom, Node nodeTo) { localFlowStep(nodeFrom, nodeTo) }
+
+private newtype TLambdaCallKind = TNone()
+
+class LambdaCallKind = TLambdaCallKind;
+
+/** Holds if `creation` is an expression that creates a lambda of kind `kind` for `c`. */
+predicate lambdaCreation(Node creation, LambdaCallKind kind, DataFlowCallable c) { none() }
+
+/** Holds if `call` is a lambda call of kind `kind` where `receiver` is the lambda expression. */
+predicate lambdaCall(DataFlowCall call, LambdaCallKind kind, Node receiver) { none() }
+
+/** Extra data-flow steps needed for lambda flow analysis. */
+predicate additionalLambdaFlowStep(Node nodeFrom, Node nodeTo, boolean preservesValue) { none() }
+
+/**
+ * Since our model is so simple, we dont want to compress the local flow steps.
+ * This compression is normally done to not show SSA steps, casts, etc.
+ */
+predicate neverSkipInPathGraph(Node node) { any() }
+
+predicate knownSourceModel(Node source, string model) { none() }
+
+predicate knownSinkModel(Node sink, string model) { none() }

actions/ql/lib/codeql/actions/dataflow/internal/DataFlowPublic.qll

@@ -0,0 +1,194 @@
+private import codeql.dataflow.DataFlow
+private import codeql.actions.Ast
+private import codeql.actions.Cfg as Cfg
+private import codeql.Locations
+private import DataFlowPrivate
+
+class Node extends TNode {
+  /** Gets a textual representation of this element. */
+  string toString() { none() }
+
+  Location getLocation() { none() }
+
+  /**
+   * Holds if this element is at the specified location.
+   * The location spans column `startcolumn` of line `startline` to
+   * column `endcolumn` of line `endline` in file `filepath`.
+   * For more information, see
+   * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
+   */
+  predicate hasLocationInfo(
+    string filepath, int startline, int startcolumn, int endline, int endcolumn
+  ) {
+    this.getLocation().hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
+  }
+
+  AstNode asExpr() { none() }
+}
+
+/**
+ * Any Ast Expression.
+ * UsesExpr, RunExpr, ArgumentExpr, VarAccessExpr, ...
+ */
+class ExprNode extends Node, TExprNode {
+  private DataFlowExpr expr;
+
+  ExprNode() { this = TExprNode(expr) }
+
+  Cfg::Node getCfgNode() { result = expr }
+
+  override string toString() { result = expr.toString() }
+
+  override Location getLocation() { result = expr.getLocation() }
+
+  override AstNode asExpr() { result = expr.getAstNode() }
+}
+
+/**
+ * Reusable workflow input nodes
+ */
+class ParameterNode extends ExprNode {
+  private Input input;
+
+  ParameterNode() { this.asExpr() = input }
+
+  predicate isParameterOf(DataFlowCallable c, ParameterPosition pos) {
+    input = c.(ReusableWorkflow).getInput(pos) or
+    input = c.(CompositeAction).getInput(pos)
+  }
+
+  override string toString() { result = "input " + input.toString() }
+
+  override Location getLocation() { result = input.getLocation() }
+
+  Input getInput() { result = input }
+}
+
+/**
+ * A call to a data flow callable (Uses).
+ */
+class CallNode extends ExprNode {
+  private DataFlowCall call;
+
+  CallNode() { this.getCfgNode() instanceof DataFlowCall }
+
+  DataFlowCallable getCalleeNode() { result = viableCallable(this.getCfgNode()) }
+}
+
+/**
+ * An argument to a Uses step (call).
+ */
+class ArgumentNode extends ExprNode {
+  ArgumentNode() { this.getCfgNode().getAstNode() = any(Uses e).getArgumentExpr(_) }
+
+  predicate argumentOf(DataFlowCall call, ArgumentPosition pos) {
+    this.getCfgNode() = call.(Cfg::Node).getASuccessor+() and
+    call.(Cfg::Node).getAstNode() =
+      any(Uses e | e.getArgumentExpr(pos) = this.getCfgNode().getAstNode())
+  }
+}
+
+/**
+ * Reusable workflow output nodes
+ */
+class ReturnNode extends ExprNode {
+  private Outputs outputs;
+
+  ReturnNode() {
+    this.asExpr() = outputs and
+    (
+      exists(ReusableWorkflow w | w.getOutputs() = outputs) or
+      exists(CompositeAction a | a.getOutputs() = outputs)
+    )
+  }
+
+  ReturnKind getKind() { result = TNormalReturn() }
+
+  override string toString() { result = "output " + outputs.toString() }
+
+  override Location getLocation() { result = outputs.getLocation() }
+}
+
+/** Gets the node corresponding to `e`. */
+Node exprNode(DataFlowExpr e) { result = TExprNode(e) }
+
+/**
+ * An entity that represents a set of `Content`s.
+ *
+ * The set may be interpreted differently depending on whether it is
+ * stored into (`getAStoreContent`) or read from (`getAReadContent`).
+ */
+class ContentSet instanceof Content {
+  /** Gets a content that may be stored into when storing into this set. */
+  Content getAStoreContent() { result = this }
+
+  /** Gets a content that may be read from when reading from this set. */
+  Content getAReadContent() { result = this }
+
+  /** Gets a textual representation of this content set. */
+  string toString() { result = super.toString() }
+
+  /**
+   * Holds if this element is at the specified location.
+   * The location spans column `startcolumn` of line `startline` to
+   * column `endcolumn` of line `endline` in file `filepath`.
+   * For more information, see
+   * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
+   */
+  predicate hasLocationInfo(
+    string filepath, int startline, int startcolumn, int endline, int endcolumn
+  ) {
+    super.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
+  }
+}
+
+/**
+ * A reference contained in an object. Examples include instance fields, the
+ * contents of a collection object, the contents of an array or pointer.
+ */
+class Content extends TContent {
+  /** Gets the type of the contained data for the purpose of type pruning. */
+  DataFlowType getType() { any() }
+
+  /** Gets a textual representation of this element. */
+  abstract string toString();
+
+  /**
+   * Holds if this element is at the specified location.
+   * The location spans column `startcolumn` of line `startline` to
+   * column `endcolumn` of line `endline` in file `filepath`.
+   * For more information, see
+   * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
+   */
+  predicate hasLocationInfo(
+    string filepath, int startline, int startcolumn, int endline, int endcolumn
+  ) {
+    filepath = "" and startline = 0 and startcolumn = 0 and endline = 0 and endcolumn = 0
+  }
+}
+
+/** A field of an object, for example an instance variable. */
+class FieldContent extends Content, TFieldContent {
+  private string name;
+
+  FieldContent() { this = TFieldContent(name) }
+
+  /** Gets the name of the field. */
+  string getName() { result = name }
+
+  override string toString() { result = name }
+}
+
+predicate hasLocalFlow(Node n1, Node n2) {
+  n1 = n2 or
+  simpleLocalFlowStep(n1, n2, _) or
+  exists(ContentSet c | ctxFieldReadStep(n1, n2, c))
+}
+
+predicate hasLocalFlowExpr(AstNode n1, AstNode n2) {
+  exists(Node dn1, Node dn2 |
+    dn1.asExpr() = n1 and
+    dn2.asExpr() = n2 and
+    hasLocalFlow(dn1, dn2)
+  )
+}

actions/ql/lib/codeql/actions/dataflow/internal/ExternalFlowExtensions.qll

@@ -0,0 +1,24 @@
+/**
+ * This module provides extensible predicates for defining MaD models.
+ */
+
+/**
+ * Holds if a source model exists for the given parameters.
+ */
+extensible predicate actionsSourceModel(
+  string action, string version, string output, string kind, string provenance
+);
+
+/**
+ * Holds if a summary model exists for the given parameters.
+ */
+extensible predicate actionsSummaryModel(
+  string action, string version, string input, string output, string kind, string provenance
+);
+
+/**
+ * Holds if a sink model exists for the given parameters.
+ */
+extensible predicate actionsSinkModel(
+  string action, string version, string input, string kind, string provenance
+);

actions/ql/lib/codeql/actions/dataflow/internal/TaintTrackingImplSpecific.qll

@@ -0,0 +1,12 @@
+/**
+ * Provides Actions-specific definitions for use in the taint tracking library.
+ * Implementation of https://github.com/github/codeql/blob/main/shared/dataflow/codeql/dataflow/TaintTracking.qll
+ */
+
+private import codeql.Locations
+private import codeql.dataflow.TaintTracking
+private import DataFlowImplSpecific
+
+module ActionsTaintTracking implements InputSig<Location, ActionsDataFlow> {
+  import TaintTrackingPrivate
+}

actions/ql/lib/codeql/actions/dataflow/internal/TaintTrackingPrivate.qll

@@ -0,0 +1,40 @@
+/**
+ * Provides modules for performing local (intra-procedural) and
+ * global (inter-procedural) taint-tracking analyses.
+ */
+
+private import DataFlowPrivate
+private import codeql.actions.DataFlow
+private import codeql.actions.dataflow.TaintSteps
+private import codeql.actions.Ast
+
+/**
+ * Holds if `node` should be a sanitizer in all global taint flow configurations
+ * but not in local taint.
+ */
+predicate defaultTaintSanitizer(DataFlow::Node node) { none() }
+
+// predicate defaultAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
+//   any(AdditionalTaintStep s).step(nodeFrom, nodeTo)
+// }
+/**
+ * Holds if the additional step from `nodeFrom` to `nodeTo` should be included
+ * in all global taint flow configurations.
+ */
+cached
+predicate defaultAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo, string model) {
+  any(AdditionalTaintStep s).step(nodeFrom, nodeTo) and model = ""
+}
+
+/**
+ * Holds if taint flow configurations should allow implicit reads of `c` at sinks
+ * and inputs to additional taint steps.
+ */
+bindingset[node]
+predicate defaultImplicitTaintRead(DataFlow::Node node, DataFlow::ContentSet c) { none() }
+
+/**
+ * Holds if the additional step from `src` to `sink` should be considered in
+ * speculative taint flow exploration.
+ */
+predicate speculativeTaintStep(DataFlow::Node src, DataFlow::Node sink) { none() }

actions/ql/lib/codeql/actions/ideContextual/IDEContextual.qll

@@ -0,0 +1,19 @@
+private import codeql.files.FileSystem
+
+/**
+ * Returns an appropriately encoded version of a filename `name`
+ * passed by the VS Code extension in order to coincide with the
+ * output of `.getFile()` on locatable entities.
+ */
+cached
+File getFileBySourceArchiveName(string name) {
+  // The name provided for a file in the source archive by the VS Code extension
+  // has some differences from the absolute path in the database:
+  // 1. colons are replaced by underscores
+  // 2. there's a leading slash, even for Windows paths: "C:/foo/bar" ->
+  //    "/C_/foo/bar"
+  // 3. double slashes in UNC prefixes are replaced with a single slash
+  // We can handle 2 and 3 together by unconditionally adding a leading slash
+  // before replacing double slashes.
+  name = ("/" + result.getAbsolutePath().replaceAll(":", "_")).replaceAll("//", "/")
+}

actions/ql/lib/codeql/actions/ideContextual/printAst.qll

@@ -0,0 +1,137 @@
+/**
+ * Provides queries to pretty-print an Kaleidoscope abstract syntax tree as a graph.
+ *
+ * By default, this will print the AST for all nodes in the database. To change
+ * this behavior, extend `PrintASTConfiguration` and override `shouldPrintNode`
+ * to hold for only the AST nodes you wish to view.
+ */
+
+private import codeql.actions.Ast
+private import codeql.Locations
+
+/**
+ * The query can extend this class to control which nodes are printed.
+ */
+class PrintAstConfiguration extends string {
+  PrintAstConfiguration() { this = "PrintAstConfiguration" }
+
+  /**
+   * Holds if the given node should be printed.
+   */
+  predicate shouldPrintNode(PrintAstNode n) { any() }
+}
+
+newtype TPrintNode = TPrintRegularAstNode(AstNode n) { any() }
+
+private predicate shouldPrintNode(PrintAstNode n) {
+  any(PrintAstConfiguration config).shouldPrintNode(n)
+}
+
+/**
+ * A node in the output tree.
+ */
+class PrintAstNode extends TPrintNode {
+  /** Gets a textual representation of this node in the PrintAst output tree. */
+  string toString() { none() }
+
+  /**
+   * Gets the child node with name `edgeName`. Typically this is the name of the
+   * predicate used to access the child.
+   */
+  PrintAstNode getChild(string edgeName) { none() }
+
+  /** Get the Location of this AST node */
+  Location getLocation() { none() }
+
+  /** Gets a child of this node. */
+  final PrintAstNode getAChild() { result = this.getChild(_) }
+
+  /** Gets the parent of this node, if any. */
+  final PrintAstNode getParent() { result.getAChild() = this }
+
+  /** Gets a value used to order this node amongst its siblings. */
+  int getOrder() {
+    this =
+      rank[result](PrintRegularAstNode p, Location l, File f |
+        l = p.getLocation() and
+        f = l.getFile()
+      |
+        p
+        order by
+          f.getBaseName(), f.getAbsolutePath(), l.getStartLine(), l.getStartColumn(),
+          l.getEndLine(), l.getEndColumn()
+      )
+  }
+
+  /**
+   * Gets the value of the property of this node, where the name of the property
+   * is `key`.
+   */
+  final string getProperty(string key) {
+    key = "semmle.label" and
+    result = this.toString()
+    or
+    key = "semmle.order" and result = this.getOrder().toString()
+  }
+}
+
+/** An `AstNode` in the output tree. */
+class PrintRegularAstNode extends PrintAstNode, TPrintRegularAstNode {
+  AstNode astNode;
+
+  PrintRegularAstNode() { this = TPrintRegularAstNode(astNode) }
+
+  override string toString() {
+    result = "[" + concat(astNode.getAPrimaryQlClass(), ", ") + "] " + astNode.toString()
+  }
+
+  override Location getLocation() { result = astNode.getLocation() }
+
+  override PrintAstNode getChild(string name) {
+    exists(int i |
+      name = i.toString() and
+      result =
+        TPrintRegularAstNode(rank[i](AstNode child, Location l |
+            child.getParentNode() = astNode and
+            child.getLocation() = l
+          |
+            child
+            order by
+              l.getStartLine(), l.getStartColumn(), l.getEndColumn(), l.getEndLine(),
+              child.toString()
+          ))
+    )
+  }
+}
+
+/**
+ * Holds if `node` belongs to the output tree, and its property `key` has the
+ * given `value`.
+ */
+query predicate nodes(PrintAstNode node, string key, string value) {
+  value = node.getProperty(key) and shouldPrintNode(node)
+}
+
+/**
+ * Holds if `target` is a child of `source` in the AST, and property `key` of
+ * the edge has the given `value`.
+ */
+query predicate edges(PrintAstNode source, PrintAstNode target, string key, string value) {
+  shouldPrintNode(source) and
+  shouldPrintNode(target) and
+  target = source.getChild(_) and
+  (
+    key = "semmle.label" and
+    value = strictconcat(string name | source.getChild(name) = target | name, "/")
+    or
+    key = "semmle.order" and
+    value = target.getProperty("semmle.order")
+  )
+}
+
+/**
+ * Holds if property `key` of the graph has the given `value`.
+ */
+query predicate graphProperties(string key, string value) {
+  key = "semmle.graphKind" and value = "tree"
+}

actions/ql/lib/codeql/actions/security/ArgumentInjectionQuery.qll

@@ -0,0 +1,94 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.DataFlow
+
+abstract class ArgumentInjectionSink extends DataFlow::Node {
+  abstract string getCommand();
+}
+
+/**
+ * Holds if a Run step declares an environment variable, uses it as the argument to a command vulnerable to argument injection.
+ * e.g.
+ *    env:
+ *      BODY: ${{ github.event.comment.body }}
+ *    run: |
+ *       sed "s/FOO/$BODY/g" > /tmp/foo
+ */
+class ArgumentInjectionFromEnvVarSink extends ArgumentInjectionSink {
+  string command;
+  string argument;
+
+  ArgumentInjectionFromEnvVarSink() {
+    exists(Run run, string var |
+      run.getScript() = this.asExpr() and
+      (
+        exists(run.getInScopeEnvVarExpr(var)) or
+        var = "GITHUB_HEAD_REF"
+      ) and
+      run.getScript().getAnEnvReachingArgumentInjectionSink(var, command, argument)
+    )
+  }
+
+  override string getCommand() { result = command }
+}
+
+/**
+ * Holds if a Run step executes a command that returns untrusted data which flows to an unsafe argument
+ * e.g.
+ *    run: |
+ *       BODY=$(git log --format=%s)
+ *       sed "s/FOO/$BODY/g" > /tmp/foo
+ */
+class ArgumentInjectionFromCommandSink extends ArgumentInjectionSink {
+  string command;
+  string argument;
+
+  ArgumentInjectionFromCommandSink() {
+    exists(CommandSource source, Run run |
+      run = source.getEnclosingRun() and
+      this.asExpr() = run.getScript() and
+      run.getScript().getACmdReachingArgumentInjectionSink(source.getCommand(), command, argument)
+    )
+  }
+
+  override string getCommand() { result = command }
+}
+
+/**
+ * Holds if a Run step declares an environment variable, uses it as the argument to a command vulnerable to argument injection.
+ */
+class ArgumentInjectionFromMaDSink extends ArgumentInjectionSink {
+  ArgumentInjectionFromMaDSink() { madSink(this, "argument-injection") }
+
+  override string getCommand() { result = "unknown" }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate a code script.
+ */
+private module ArgumentInjectionConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source instanceof RemoteFlowSource
+    or
+    exists(Run run |
+      run.getScript() = source.asExpr() and
+      run.getScript().getAnEnvReachingArgumentInjectionSink("GITHUB_HEAD_REF", _, _)
+    )
+  }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof ArgumentInjectionSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run, string var |
+      run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+      succ.asExpr() = run.getScript() and
+      run.getScript().getAnEnvReachingArgumentInjectionSink(var, _, _)
+    )
+  }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate a code script. */
+module ArgumentInjectionFlow = TaintTracking::Global<ArgumentInjectionConfig>;

actions/ql/lib/codeql/actions/security/ArtifactPoisoningQuery.qll

@@ -0,0 +1,322 @@
+import actions
+private import codeql.actions.TaintTracking
+import codeql.actions.DataFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.security.PoisonableSteps
+import codeql.actions.security.UntrustedCheckoutQuery
+
+string unzipRegexp() { result = "(unzip|tar)\\s+.*" }
+
+string unzipDirArgRegexp() { result = "(-d|-C)\\s+([^ ]+).*" }
+
+abstract class UntrustedArtifactDownloadStep extends Step {
+  abstract string getPath();
+}
+
+class GitHubDownloadArtifactActionStep extends UntrustedArtifactDownloadStep, UsesStep {
+  GitHubDownloadArtifactActionStep() {
+    this.getCallee() = "actions/download-artifact" and
+    (
+      // By default, the permissions are scoped so they can only download Artifacts within the current workflow run.
+      // To elevate permissions for this scenario, you can specify a github-token along with other repository and run identifiers
+      this.getArgument("run-id").matches("%github.event.workflow_run.id%") and
+      exists(this.getArgument("github-token"))
+      or
+      // There is an artifact upload step in the same workflow which can be influenced by an attacker on a checkout step
+      exists(LocalJob job, SimplePRHeadCheckoutStep checkout, UsesStep upload |
+        this.getEnclosingWorkflow().getAJob() = job and
+        job.getAStep() = checkout and
+        checkout.getATriggerEvent().getName() = "pull_request_target" and
+        checkout.getAFollowingStep() = upload and
+        upload.getCallee() = "actions/upload-artifact"
+      )
+    )
+  }
+
+  override string getPath() {
+    if exists(this.getArgument("path"))
+    then result = normalizePath(this.getArgument("path"))
+    else result = "GITHUB_WORKSPACE/"
+  }
+}
+
+class DownloadArtifactActionStep extends UntrustedArtifactDownloadStep, UsesStep {
+  DownloadArtifactActionStep() {
+    this.getCallee() =
+      [
+        "dawidd6/action-download-artifact", "marcofaggian/action-download-multiple-artifacts",
+        "benday-inc/download-latest-artifact", "blablacar/action-download-last-artifact",
+        "levonet/action-download-last-artifact", "bettermarks/action-artifact-download",
+        "aochmann/actions-download-artifact", "cytopia/download-artifact-retry-action",
+        "alextompkins/download-prior-artifact", "nmerget/download-gzip-artifact",
+        "benday-inc/download-artifact", "synergy-au/download-workflow-artifacts-action",
+        "ishworkh/docker-image-artifact-download", "ishworkh/container-image-artifact-download",
+        "sidx1024/action-download-artifact", "hyperskill/azblob-download-artifact",
+        "ma-ve/action-download-artifact-with-retry"
+      ] and
+    (
+      not exists(this.getArgument(["branch", "branch_name"]))
+      or
+      exists(this.getArgument(["branch", "branch_name"])) and
+      this.getArgument("allow_forks") = "true"
+    ) and
+    (
+      not exists(this.getArgument(["commit", "commitHash", "commit_sha"])) or
+      not this.getArgument(["commit", "commitHash", "commit_sha"])
+          .matches("%github.event.pull_request.head.sha%")
+    ) and
+    (
+      not exists(this.getArgument("event")) or
+      not this.getArgument("event") = "pull_request"
+    ) and
+    (
+      not exists(this.getArgument(["run-id", "run_id", "workflow-run-id", "workflow_run_id"])) or
+      this.getArgument(["run-id", "run_id", "workflow-run-id", "workflow_run_id"])
+          .matches("%github.event.workflow_run.id%")
+    ) and
+    (
+      not exists(this.getArgument("pr")) or
+      not this.getArgument("pr")
+          .matches(["%github.event.pull_request.number%", "%github.event.number%"])
+    )
+  }
+
+  override string getPath() {
+    if exists(this.getArgument(["path", "download_path"]))
+    then result = normalizePath(this.getArgument(["path", "download_path"]))
+    else
+      if exists(this.getArgument("paths"))
+      then result = normalizePath(this.getArgument("paths").splitAt(" "))
+      else result = "GITHUB_WORKSPACE/"
+  }
+}
+
+class LegitLabsDownloadArtifactActionStep extends UntrustedArtifactDownloadStep, UsesStep {
+  LegitLabsDownloadArtifactActionStep() {
+    this.getCallee() = "Legit-Labs/action-download-artifact" and
+    (
+      not exists(this.getArgument("branch")) or
+      not this.getArgument("branch") = ["main", "master"]
+    ) and
+    (
+      not exists(this.getArgument("commit")) or
+      not this.getArgument("commit").matches("%github.event.pull_request.head.sha%")
+    ) and
+    (
+      not exists(this.getArgument("event")) or
+      not this.getArgument("event") = "pull_request"
+    ) and
+    (
+      not exists(this.getArgument("run_id")) or
+      not this.getArgument("run_id").matches("%github.event.workflow_run.id%")
+    ) and
+    (
+      not exists(this.getArgument("pr")) or
+      not this.getArgument("pr").matches("%github.event.pull_request.number%")
+    )
+  }
+
+  override string getPath() {
+    if exists(this.getArgument("path"))
+    then result = normalizePath(this.getArgument("path"))
+    else result = "GITHUB_WORKSPACE/artifacts"
+  }
+}
+
+class ActionsGitHubScriptDownloadStep extends UntrustedArtifactDownloadStep, UsesStep {
+  string script;
+
+  ActionsGitHubScriptDownloadStep() {
+    // eg:
+    // - uses: actions/github-script@v6
+    //   with:
+    //     script: |
+    //       let allArtifacts = await github.rest.actions.listWorkflowRunArtifacts({
+    //           owner: context.repo.owner,
+    //           repo: context.repo.repo,
+    //           run_id: context.payload.workflow_run.id,
+    //       });
+    //       let matchArtifact = allArtifacts.data.artifacts.filter((artifact) => {
+    //           return artifact.name == "<ARTEFACT_NAME>"
+    //       })[0];
+    //       let download = await github.rest.actions.downloadArtifact({
+    //           owner: context.repo.owner,
+    //           repo: context.repo.repo,
+    //           artifact_id: matchArtifact.id,
+    //           archive_format: 'zip',
+    //       });
+    //      var fs = require('fs');
+    //      fs.writeFileSync('${{github.workspace}}/test-results.zip', Buffer.from(download.data));
+    this.getCallee() = "actions/github-script" and
+    this.getArgument("script") = script and
+    script.matches("%listWorkflowRunArtifacts(%") and
+    script.matches("%downloadArtifact(%") and
+    script.matches("%writeFileSync(%") and
+    // Filter out artifacts that were created by pull-request.
+    not script.matches("%exclude_pull_requests: true%")
+  }
+
+  override string getPath() {
+    if
+      this.getAFollowingStep()
+          .(Run)
+          .getScript()
+          .getACommand()
+          .regexpMatch(unzipRegexp() + unzipDirArgRegexp())
+    then
+      result =
+        normalizePath(trimQuotes(this.getAFollowingStep()
+                .(Run)
+                .getScript()
+                .getACommand()
+                .regexpCapture(unzipRegexp() + unzipDirArgRegexp(), 3)))
+    else
+      if this.getAFollowingStep().(Run).getScript().getACommand().regexpMatch(unzipRegexp())
+      then result = "GITHUB_WORKSPACE/"
+      else none()
+  }
+}
+
+class GHRunArtifactDownloadStep extends UntrustedArtifactDownloadStep, Run {
+  GHRunArtifactDownloadStep() {
+    // eg: - run: gh run download ${{ github.event.workflow_run.id }} --repo "${GITHUB_REPOSITORY}" --name "artifact_name"
+    this.getScript().getACommand().regexpMatch(".*gh\\s+run\\s+download.*") and
+    (
+      this.getScript().getACommand().regexpMatch(unzipRegexp()) or
+      this.getAFollowingStep().(Run).getScript().getACommand().regexpMatch(unzipRegexp())
+    )
+  }
+
+  override string getPath() {
+    if
+      this.getAFollowingStep()
+          .(Run)
+          .getScript()
+          .getACommand()
+          .regexpMatch(unzipRegexp() + unzipDirArgRegexp()) or
+      this.getScript().getACommand().regexpMatch(unzipRegexp() + unzipDirArgRegexp())
+    then
+      result =
+        normalizePath(trimQuotes(this.getScript()
+                .getACommand()
+                .regexpCapture(unzipRegexp() + unzipDirArgRegexp(), 3))) or
+      result =
+        normalizePath(trimQuotes(this.getAFollowingStep()
+                .(Run)
+                .getScript()
+                .getACommand()
+                .regexpCapture(unzipRegexp() + unzipDirArgRegexp(), 3)))
+    else
+      if
+        this.getAFollowingStep().(Run).getScript().getACommand().regexpMatch(unzipRegexp()) or
+        this.getScript().getACommand().regexpMatch(unzipRegexp())
+      then result = "GITHUB_WORKSPACE/"
+      else none()
+  }
+}
+
+class DirectArtifactDownloadStep extends UntrustedArtifactDownloadStep, Run {
+  DirectArtifactDownloadStep() {
+    // eg:
+    // run: |
+    //   artifacts_url=${{ github.event.workflow_run.artifacts_url }}
+    //   gh api "$artifacts_url" -q '.artifacts[] | [.name, .archive_download_url] | @tsv' | while read artifact
+    //   do
+    //     IFS=$'\t' read name url <<< "$artifact"
+    //     gh api $url > "$name.zip"
+    //     unzip -d "$name" "$name.zip"
+    //   done
+    this.getScript().getACommand().matches("%github.event.workflow_run.artifacts_url%") and
+    (
+      this.getScript().getACommand().regexpMatch(unzipRegexp()) or
+      this.getAFollowingStep().(Run).getScript().getACommand().regexpMatch(unzipRegexp())
+    )
+  }
+
+  override string getPath() {
+    if
+      this.getScript().getACommand().regexpMatch(unzipRegexp() + unzipDirArgRegexp()) or
+      this.getAFollowingStep()
+          .(Run)
+          .getScript()
+          .getACommand()
+          .regexpMatch(unzipRegexp() + unzipDirArgRegexp())
+    then
+      result =
+        normalizePath(trimQuotes(this.getScript()
+                .getACommand()
+                .regexpCapture(unzipRegexp() + unzipDirArgRegexp(), 3))) or
+      result =
+        normalizePath(trimQuotes(this.getAFollowingStep()
+                .(Run)
+                .getScript()
+                .getACommand()
+                .regexpCapture(unzipRegexp() + unzipDirArgRegexp(), 3)))
+    else result = "GITHUB_WORKSPACE/"
+  }
+}
+
+class ArtifactPoisoningSink extends DataFlow::Node {
+  UntrustedArtifactDownloadStep download;
+  PoisonableStep poisonable;
+
+  ArtifactPoisoningSink() {
+    download.getAFollowingStep() = poisonable and
+    // excluding artifacts downloaded to /tmp
+    not download.getPath().regexpMatch("^/tmp.*") and
+    (
+      poisonable.(Run).getScript() = this.asExpr() and
+      (
+        // Check if the poisonable step is a local script execution step
+        // and the path of the command or script matches the path of the downloaded artifact
+        isSubpath(poisonable.(LocalScriptExecutionRunStep).getPath(), download.getPath())
+        or
+        // Checking the path for non local script execution steps is very difficult
+        not poisonable instanceof LocalScriptExecutionRunStep
+        // Its not easy to extract the path from a non-local script execution step so skipping this check for now
+        // and isSubpath(poisonable.(Run).getWorkingDirectory(), download.getPath())
+      )
+      or
+      poisonable.(UsesStep) = this.asExpr() and
+      (
+        not poisonable instanceof LocalActionUsesStep and
+        download.getPath() = "GITHUB_WORKSPACE/"
+        or
+        isSubpath(poisonable.(LocalActionUsesStep).getPath(), download.getPath())
+      )
+    )
+  }
+
+  string getPath() { result = download.getPath() }
+}
+
+/**
+ * A taint-tracking configuration for unsafe artifacts
+ * that is used may lead to artifact poisoning
+ */
+private module ArtifactPoisoningConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof ArtifactSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof ArtifactPoisoningSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(PoisonableStep step |
+      pred instanceof ArtifactSource and
+      pred.asExpr().(Step).getAFollowingStep() = step and
+      (
+        succ.asExpr() = step.(Run).getScript() or
+        succ.asExpr() = step.(UsesStep)
+      )
+    )
+    or
+    exists(Run run |
+      pred instanceof ArtifactSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+/** Tracks flow of unsafe artifacts that is used in an insecure way. */
+module ArtifactPoisoningFlow = TaintTracking::Global<ArtifactPoisoningConfig>;

actions/ql/lib/codeql/actions/security/CachePoisoningQuery.qll

@@ -0,0 +1,72 @@
+import actions
+
+string defaultBranchTriggerEvent() {
+  result =
+    [
+      "check_run", "check_suite", "delete", "discussion", "discussion_comment", "fork", "gollum",
+      "issue_comment", "issues", "label", "milestone", "project", "project_card", "project_column",
+      "public", "pull_request_comment", "pull_request_target", "repository_dispatch", "schedule",
+      "watch", "workflow_run"
+    ]
+}
+
+predicate runsOnDefaultBranch(Event e) {
+  (
+    e.getName() = defaultBranchTriggerEvent() and
+    not e.getName() = "pull_request_target"
+    or
+    e.getName() = "push" and
+    e.getAPropertyValue("branches") = defaultBranchNames()
+    or
+    e.getName() = "pull_request_target" and
+    (
+      // no filtering
+      not e.hasProperty("branches") and not e.hasProperty("branches-ignore")
+      or
+      // only branches-ignore filter
+      e.hasProperty("branches-ignore") and
+      not e.hasProperty("branches") and
+      not e.getAPropertyValue("branches-ignore") = defaultBranchNames()
+      or
+      // only branches filter
+      e.hasProperty("branches") and
+      not e.hasProperty("branches-ignore") and
+      e.getAPropertyValue("branches") = defaultBranchNames()
+      or
+      // branches and branches-ignore filters
+      e.hasProperty("branches") and
+      e.hasProperty("branches-ignore") and
+      e.getAPropertyValue("branches") = defaultBranchNames() and
+      not e.getAPropertyValue("branches-ignore") = defaultBranchNames()
+    )
+  )
+}
+
+abstract class CacheWritingStep extends Step {
+  abstract string getPath();
+}
+
+class CacheActionUsesStep extends CacheWritingStep, UsesStep {
+  CacheActionUsesStep() { this.getCallee() = "actions/cache" }
+
+  override string getPath() {
+    result = normalizePath(this.(UsesStep).getArgument("path").splitAt("\n"))
+  }
+}
+
+class CacheActionSaveUsesStep extends CacheWritingStep, UsesStep {
+  CacheActionSaveUsesStep() { this.getCallee() = "actions/cache/save" }
+
+  override string getPath() {
+    result = normalizePath(this.(UsesStep).getArgument("path").splitAt("\n"))
+  }
+}
+
+class SetupRubyUsesStep extends CacheWritingStep, UsesStep {
+  SetupRubyUsesStep() {
+    this.getCallee() = ["actions/setup-ruby", "ruby/setup-ruby"] and
+    this.getArgument("bundler-cache") = "true"
+  }
+
+  override string getPath() { result = normalizePath("vendor/bundle") }
+}

actions/ql/lib/codeql/actions/security/CodeInjectionQuery.qll

@@ -0,0 +1,41 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.DataFlow
+
+class CodeInjectionSink extends DataFlow::Node {
+  CodeInjectionSink() {
+    exists(Run e | e.getAnScriptExpr() = this.asExpr()) or
+    madSink(this, "code-injection")
+  }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate a code script.
+ */
+private module CodeInjectionConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof CodeInjectionSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Uses step |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = step and
+      succ.asExpr() = step and
+      madSink(succ, "code-injection")
+    )
+    or
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate a code script. */
+module CodeInjectionFlow = TaintTracking::Global<CodeInjectionConfig>;

actions/ql/lib/codeql/actions/security/CommandInjectionQuery.qll

@@ -0,0 +1,22 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.DataFlow
+
+private class CommandInjectionSink extends DataFlow::Node {
+  CommandInjectionSink() { madSink(this, "command-injection") }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate a system command.
+ */
+private module CommandInjectionConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof CommandInjectionSink }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate a system command. */
+module CommandInjectionFlow = TaintTracking::Global<CommandInjectionConfig>;

actions/ql/lib/codeql/actions/security/ControlChecks.qll

@@ -0,0 +1,312 @@
+import actions
+
+string any_category() {
+  result =
+    [
+      "untrusted-checkout", "output-clobbering", "envpath-injection", "envvar-injection",
+      "command-injection", "argument-injection", "code-injection", "cache-poisoning",
+      "untrusted-checkout-toctou", "artifact-poisoning", "artifact-poisoning-toctou"
+    ]
+}
+
+string non_toctou_category() {
+  result = any_category() and not result = "untrusted-checkout-toctou"
+}
+
+string toctou_category() { result = ["untrusted-checkout-toctou", "artifact-poisoning-toctou"] }
+
+string any_event() { result = actor_not_attacker_event() or result = actor_is_attacker_event() }
+
+string actor_is_attacker_event() {
+  result =
+    [
+      // actor and attacker have to be the same
+      "pull_request_target",
+      "workflow_run",
+      "discussion_comment",
+      "discussion",
+      "issues",
+      "fork",
+      "watch"
+    ]
+}
+
+string actor_not_attacker_event() {
+  result =
+    [
+      // actor and attacker can be different
+      // actor may be a collaborator, but the attacker is may be the author of the PR that gets commented
+      // therefore it may be vulnerable to TOCTOU races where the actor reviews one thing and the attacker changes it
+      "issue_comment",
+      "pull_request_comment",
+    ]
+}
+
+/** An If node that contains an actor, user or label check */
+abstract class ControlCheck extends AstNode {
+  ControlCheck() {
+    this instanceof If or
+    this instanceof Environment or
+    this instanceof UsesStep or
+    this instanceof Run
+  }
+
+  predicate protects(AstNode node, Event event, string category) {
+    // The check dominates the step it should protect
+    this.dominates(node) and
+    // The check is effective against the event and category
+    this.protectsCategoryAndEvent(category, event.getName()) and
+    // The check can be triggered by the event
+    this.getATriggerEvent() = event
+  }
+
+  predicate dominates(AstNode node) {
+    this instanceof If and
+    (
+      node.getEnclosingStep().getIf() = this or
+      node.getEnclosingJob().getIf() = this or
+      node.getEnclosingJob().getANeededJob().(LocalJob).getAStep().getIf() = this or
+      node.getEnclosingJob().getANeededJob().(LocalJob).getIf() = this
+    )
+    or
+    this instanceof Environment and
+    (
+      node.getEnclosingJob().getEnvironment() = this
+      or
+      node.getEnclosingJob().getANeededJob().getEnvironment() = this
+    )
+    or
+    (
+      this instanceof Run or
+      this instanceof UsesStep
+    ) and
+    (
+      this.(Step).getAFollowingStep() = node.getEnclosingStep()
+      or
+      node.getEnclosingJob().getANeededJob().(LocalJob).getAStep() = this.(Step)
+    )
+  }
+
+  abstract predicate protectsCategoryAndEvent(string category, string event);
+}
+
+abstract class AssociationCheck extends ControlCheck {
+  // Checks if the actor is a MEMBER/OWNER the repo
+  // - they are effective against pull requests and workflow_run (since these are triggered by pull_requests) since they can control who is making the PR
+  // - they are not effective against issue_comment since the author of the comment may not be the same as the author of the PR
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = actor_is_attacker_event() and category = any_category()
+    or
+    event = actor_not_attacker_event() and category = non_toctou_category()
+  }
+}
+
+abstract class ActorCheck extends ControlCheck {
+  // checks for a specific actor
+  // - they are effective against pull requests and workflow_run (since these are triggered by pull_requests) since they can control who is making the PR
+  // - they are not effective against issue_comment since the author of the comment may not be the same as the author of the PR
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = actor_is_attacker_event() and category = any_category()
+    or
+    event = actor_not_attacker_event() and category = non_toctou_category()
+  }
+}
+
+abstract class RepositoryCheck extends ControlCheck {
+  // checks that the origin of the code is the same as the repository.
+  // for pull_requests, that means that it triggers only on local branches or repos from the same org
+  // - they are effective against pull requests/workflow_run since they can control where the code is coming from
+  // - they are not effective against issue_comment since the repository will always be the same
+}
+
+abstract class PermissionCheck extends ControlCheck {
+  // checks that the actor has a specific permission level
+  // - they are effective against pull requests/workflow_run since they can control who can make changes
+  // - they are not effective against issue_comment since the author of the comment may not be the same as the author of the PR
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = actor_is_attacker_event() and category = any_category()
+    or
+    event = actor_not_attacker_event() and category = non_toctou_category()
+  }
+}
+
+abstract class LabelCheck extends ControlCheck {
+  // checks if the issue/pull_request is labeled, which implies that it could have been approved
+  // - they dont protect against mutation attacks
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = actor_is_attacker_event() and category = any_category()
+    or
+    event = actor_not_attacker_event() and category = non_toctou_category()
+  }
+}
+
+class EnvironmentCheck extends ControlCheck instanceof Environment {
+  // Environment checks are not effective against any mutable attacks
+  // they do actually protect against untrusted code execution (sha)
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = actor_is_attacker_event() and category = any_category()
+    or
+    event = actor_not_attacker_event() and category = non_toctou_category()
+  }
+}
+
+abstract class CommentVsHeadDateCheck extends ControlCheck {
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    // by itself, this check is not effective against any attacks
+    event = actor_not_attacker_event() and category = toctou_category()
+  }
+}
+
+/* Specific implementations of control checks */
+class LabelIfCheck extends LabelCheck instanceof If {
+  string condition;
+
+  LabelIfCheck() {
+    condition = normalizeExpr(this.getCondition()) and
+    (
+      // eg: contains(github.event.pull_request.labels.*.name, 'safe to test')
+      condition.regexpMatch(".*(^|[^!])contains\\(\\s*github\\.event\\.pull_request\\.labels\\b.*")
+      or
+      // eg: github.event.label.name == 'safe to test'
+      condition.regexpMatch(".*\\bgithub\\.event\\.label\\.name\\s*==.*")
+    )
+  }
+}
+
+class ActorIfCheck extends ActorCheck instanceof If {
+  ActorIfCheck() {
+    // eg: github.event.pull_request.user.login == 'admin'
+    exists(
+      normalizeExpr(this.getCondition())
+          .regexpFind([
+              "\\bgithub\\.event\\.pull_request\\.user\\.login\\b",
+              "\\bgithub\\.event\\.head_commit\\.author\\.name\\b",
+              "\\bgithub\\.event\\.commits.*\\.author\\.name\\b",
+              "\\bgithub\\.event\\.sender\\.login\\b"
+            ], _, _)
+    )
+    or
+    // eg: github.actor == 'admin'
+    // eg: github.triggering_actor == 'admin'
+    exists(
+      normalizeExpr(this.getCondition())
+          .regexpFind(["\\bgithub\\.actor\\b", "\\bgithub\\.triggering_actor\\b",], _, _)
+    ) and
+    not normalizeExpr(this.getCondition()).matches("%[bot]%")
+  }
+}
+
+class PullRequestTargetRepositoryIfCheck extends RepositoryCheck instanceof If {
+  PullRequestTargetRepositoryIfCheck() {
+    // eg: github.event.pull_request.head.repo.full_name == github.repository
+    exists(
+      normalizeExpr(this.getCondition())
+          // github.repository in a workflow_run event triggered by a pull request is the base repository
+          .regexpFind([
+              "\\bgithub\\.repository\\b", "\\bgithub\\.repository_owner\\b",
+              "\\bgithub\\.event\\.pull_request\\.head\\.repo\\.full_name\\b",
+              "\\bgithub\\.event\\.pull_request\\.head\\.repo\\.owner\\.name\\b",
+              "\\bgithub\\.event\\.workflow_run\\.head_repository\\.full_name\\b",
+              "\\bgithub\\.event\\.workflow_run\\.head_repository\\.owner\\.name\\b"
+            ], _, _)
+    )
+  }
+
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = "pull_request_target" and category = any_category()
+  }
+}
+
+class WorkflowRunRepositoryIfCheck extends RepositoryCheck instanceof If {
+  WorkflowRunRepositoryIfCheck() {
+    // eg: github.event.workflow_run.head_repository.full_name == github.repository
+    exists(
+      normalizeExpr(this.getCondition())
+          // github.repository in a workflow_run event triggered by a pull request is the base repository
+          .regexpFind([
+              "\\bgithub\\.event\\.workflow_run\\.head_repository\\.full_name\\b",
+              "\\bgithub\\.event\\.workflow_run\\.head_repository\\.owner\\.name\\b"
+            ], _, _)
+    )
+  }
+
+  override predicate protectsCategoryAndEvent(string category, string event) {
+    event = "workflow_run" and category = any_category()
+  }
+}
+
+class AssociationIfCheck extends AssociationCheck instanceof If {
+  AssociationIfCheck() {
+    // eg: contains(fromJson('["MEMBER", "OWNER"]'), github.event.comment.author_association)
+    normalizeExpr(this.getCondition())
+        .splitAt("\n")
+        .regexpMatch([
+            ".*\\bgithub\\.event\\.comment\\.author_association\\b.*",
+            ".*\\bgithub\\.event\\.issue\\.author_association\\b.*",
+            ".*\\bgithub\\.event\\.pull_request\\.author_association\\b.*",
+          ])
+  }
+}
+
+class AssociationActionCheck extends AssociationCheck instanceof UsesStep {
+  AssociationActionCheck() {
+    this.getCallee() = "TheModdingInquisition/actions-team-membership" and
+    (
+      not exists(this.getArgument("exit"))
+      or
+      this.getArgument("exit") = "true"
+    )
+    or
+    this.getCallee() = "actions/github-script" and
+    this.getArgument("script").splitAt("\n").matches("%getMembershipForUserInOrg%")
+    or
+    this.getCallee() = "octokit/request-action" and
+    this.getArgument("route").regexpMatch("GET.*(memberships).*")
+  }
+}
+
+class PermissionActionCheck extends PermissionCheck instanceof UsesStep {
+  PermissionActionCheck() {
+    this.getCallee() = "actions-cool/check-user-permission" and
+    (
+      // default permission level is write
+      not exists(this.getArgument("permission-level")) or
+      this.getArgument("require") = ["write", "admin"]
+    )
+    or
+    this.getCallee() = "sushichop/action-repository-permission" and
+    this.getArgument("required-permission") = ["write", "admin"]
+    or
+    this.getCallee() = "prince-chrismc/check-actor-permissions-action" and
+    this.getArgument("permission") = ["write", "admin"]
+    or
+    this.getCallee() = "lannonbr/repo-permission-check-action" and
+    this.getArgument("permission") = ["write", "admin"]
+    or
+    this.getCallee() = "xt0rted/slash-command-action" and
+    (
+      // default permission level is write
+      not exists(this.getArgument("permission-level")) or
+      this.getArgument("permission-level") = ["write", "admin"]
+    )
+    or
+    this.getCallee() = "actions/github-script" and
+    this.getArgument("script").splitAt("\n").matches("%getCollaboratorPermissionLevel%")
+    or
+    this.getCallee() = "octokit/request-action" and
+    this.getArgument("route").regexpMatch("GET.*(collaborators|permission).*")
+  }
+}
+
+class BashCommentVsHeadDateCheck extends CommentVsHeadDateCheck, Run {
+  BashCommentVsHeadDateCheck() {
+    // eg: if [[ $(date -d "$pushed_at" +%s) -gt $(date -d "$COMMENT_AT" +%s) ]]; then
+    exists(string cmd1, string cmd2 |
+      cmd1 = this.getScript().getACommand() and
+      cmd2 = this.getScript().getACommand() and
+      not cmd1 = cmd2 and
+      cmd1.toLowerCase().regexpMatch("date\\s+-d.*(commit|pushed|comment|commented)_at.*") and
+      cmd2.toLowerCase().regexpMatch("date\\s+-d.*(commit|pushed|comment|commented)_at.*")
+    )
+  }
+}

actions/ql/lib/codeql/actions/security/EnvPathInjectionQuery.qll

@@ -0,0 +1,114 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+private import codeql.actions.security.ArtifactPoisoningQuery
+private import codeql.actions.security.UntrustedCheckoutQuery
+
+abstract class EnvPathInjectionSink extends DataFlow::Node { }
+
+/**
+ * Holds if a Run step declares a PATH environment variable with contents from a local file.
+ */
+class EnvPathInjectionFromFileReadSink extends EnvPathInjectionSink {
+  EnvPathInjectionFromFileReadSink() {
+    exists(Run run, Step step |
+      (
+        step instanceof UntrustedArtifactDownloadStep or
+        step instanceof PRHeadCheckoutStep
+      ) and
+      this.asExpr() = run.getScript() and
+      step.getAFollowingStep() = run and
+      (
+        // echo "$(cat foo.txt)" >> $GITHUB_PATH
+        // FOO=$(cat foo.txt)
+        // echo "$FOO" >> $GITHUB_PATH
+        exists(string cmd |
+          run.getScript().getAFileReadCommand() = cmd and
+          run.getScript().getACmdReachingGitHubPathWrite(cmd)
+        )
+        or
+        // cat foo.txt >> $GITHUB_PATH
+        run.getScript().fileToGitHubPath(_)
+      )
+    )
+  }
+}
+
+/**
+ * Holds if a Run step executes a command that returns untrusted data which flows to GITHUB_ENV
+ * e.g.
+ *    run: |
+ *          COMMIT_MESSAGE=$(git log --format=%s)
+ *          echo "${COMMIT_MESSAGE}" >> $GITHUB_PATH
+ */
+class EnvPathInjectionFromCommandSink extends EnvPathInjectionSink {
+  EnvPathInjectionFromCommandSink() {
+    exists(CommandSource source |
+      this.asExpr() = source.getEnclosingRun().getScript() and
+      source.getEnclosingRun().getScript().getACmdReachingGitHubPathWrite(source.getCommand())
+    )
+  }
+}
+
+/**
+ * Holds if a Run step declares an environment variable, uses it to declare a PATH env var.
+ * e.g.
+ *    env:
+ *      BODY: ${{ github.event.comment.body }}
+ *    run: |
+ *      echo "$BODY" >> $GITHUB_PATH
+ */
+class EnvPathInjectionFromEnvVarSink extends EnvPathInjectionSink {
+  EnvPathInjectionFromEnvVarSink() {
+    exists(Run run, string var_name |
+      run.getScript().getAnEnvReachingGitHubPathWrite(var_name) and
+      exists(run.getInScopeEnvVarExpr(var_name)) and
+      run.getScript() = this.asExpr()
+    )
+  }
+}
+
+class EnvPathInjectionFromMaDSink extends EnvPathInjectionSink {
+  EnvPathInjectionFromMaDSink() { madSink(this, "envpath-injection") }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate an environment variable.
+ */
+private module EnvPathInjectionConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof EnvPathInjectionSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run, string var |
+      run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+      succ.asExpr() = run.getScript() and
+      (
+        run.getScript().getAnEnvReachingGitHubEnvWrite(var, _)
+        or
+        run.getScript().getAnEnvReachingGitHubOutputWrite(var, _)
+        or
+        run.getScript().getAnEnvReachingGitHubPathWrite(var)
+      )
+    )
+    or
+    exists(Uses step |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = step and
+      succ.asExpr() = step and
+      madSink(succ, "envpath-injection")
+    )
+    or
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate the PATH environment variable. */
+module EnvPathInjectionFlow = TaintTracking::Global<EnvPathInjectionConfig>;

actions/ql/lib/codeql/actions/security/EnvVarInjectionQuery.qll

@@ -0,0 +1,169 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+private import codeql.actions.security.ArtifactPoisoningQuery
+private import codeql.actions.security.UntrustedCheckoutQuery
+
+abstract class EnvVarInjectionSink extends DataFlow::Node { }
+
+string sanitizerCommand() {
+  result =
+    [
+      "tr\\s+(-d\\s*)?('|\")?.n('|\")?", // tr -d '\n' ' ', tr '\n' ' '
+      "tr\\s+-cd\\s+.*:al(pha|num):", // tr -cd '[:alpha:_]'
+      "(head|tail)\\s+-n\\s+1" // head -n 1, tail -n 1
+    ]
+}
+
+/**
+ * Holds if a Run step declares an environment variable with contents from a local file.
+ */
+class EnvVarInjectionFromFileReadSink extends EnvVarInjectionSink {
+  EnvVarInjectionFromFileReadSink() {
+    exists(Run run, Step step |
+      (
+        step instanceof UntrustedArtifactDownloadStep or
+        step instanceof PRHeadCheckoutStep
+      ) and
+      this.asExpr() = run.getScript() and
+      step.getAFollowingStep() = run and
+      (
+        // eg:
+        // echo "SHA=$(cat test-results/sha-number)" >> $GITHUB_ENV
+        // echo "SHA=$(<test-results/sha-number)" >> $GITHUB_ENV
+        // FOO=$(cat test-results/sha-number)
+        // echo "FOO=$FOO" >> $GITHUB_ENV
+        exists(string cmd, string var, string sanitizer |
+          run.getScript().getAFileReadCommand() = cmd and
+          run.getScript().getACmdReachingGitHubEnvWrite(cmd, var) and
+          run.getScript().getACmdReachingGitHubEnvWrite(sanitizer, var) and
+          not exists(sanitizer.regexpFind(sanitizerCommand(), _, _))
+        )
+        or
+        // eg: cat test-results/.env >> $GITHUB_ENV
+        run.getScript().fileToGitHubEnv(_)
+      )
+    )
+  }
+}
+
+/**
+ * Holds if a Run step executes a command that returns untrusted data which flows to GITHUB_ENV
+ * e.g.
+ *    run: |
+ *          COMMIT_MESSAGE=$(git log --format=%s)
+ *          echo "COMMIT_MESSAGE=${COMMIT_MESSAGE}" >> $GITHUB_ENV
+ */
+class EnvVarInjectionFromCommandSink extends EnvVarInjectionSink {
+  CommandSource inCommand;
+  string injectedVar;
+  string command;
+
+  EnvVarInjectionFromCommandSink() {
+    exists(Run run |
+      this.asExpr() = inCommand.getEnclosingRun().getScript() and
+      run = inCommand.getEnclosingRun() and
+      run.getScript().getACmdReachingGitHubEnvWrite(inCommand.getCommand(), injectedVar) and
+      (
+        // the source flows to the injected variable without any command in between
+        not run.getScript().getACmdReachingGitHubEnvWrite(_, injectedVar) and
+        command = ""
+        or
+        // the source flows to the injected variable with a command in between
+        run.getScript().getACmdReachingGitHubEnvWrite(command, injectedVar) and
+        not command.regexpMatch(".*" + sanitizerCommand() + ".*")
+      )
+    )
+  }
+}
+
+/**
+ * Holds if a Run step declares an environment variable, uses it to declare env var.
+ * e.g.
+ *    env:
+ *      BODY: ${{ github.event.comment.body }}
+ *    run: |
+ *      echo "FOO=$BODY" >> $GITHUB_ENV
+ */
+class EnvVarInjectionFromEnvVarSink extends EnvVarInjectionSink {
+  string inVar;
+  string injectedVar;
+  string command;
+
+  EnvVarInjectionFromEnvVarSink() {
+    exists(Run run |
+      run.getScript() = this.asExpr() and
+      exists(run.getInScopeEnvVarExpr(inVar)) and
+      run.getScript().getAnEnvReachingGitHubEnvWrite(inVar, injectedVar) and
+      (
+        // the source flows to the injected variable without any command in between
+        not run.getScript().getACmdReachingGitHubEnvWrite(_, injectedVar) and
+        command = ""
+        or
+        // the source flows to the injected variable with a command in between
+        run.getScript().getACmdReachingGitHubEnvWrite(_, injectedVar) and
+        run.getScript().getACmdReachingGitHubEnvWrite(command, injectedVar) and
+        not command.regexpMatch(".*" + sanitizerCommand() + ".*")
+      )
+    )
+  }
+}
+
+/**
+ * Holds if a 3rd party action declares an environment variable with contents from an untrusted file.
+ * e.g.
+ *- name: Load .env file
+ *  uses: aarcangeli/load-dotenv@v1.0.0
+ *  with:
+ *    path: 'backend/new'
+ *    filenames: |
+ *      .env
+ *      .env.test
+ *    quiet: false
+ *    if-file-not-found: error
+ */
+class EnvVarInjectionFromMaDSink extends EnvVarInjectionSink {
+  EnvVarInjectionFromMaDSink() { madSink(this, "envvar-injection") }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate an environment variable.
+ */
+private module EnvVarInjectionConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source instanceof RemoteFlowSource and
+    not source.(RemoteFlowSource).getSourceType() = ["branch", "username"]
+  }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof EnvVarInjectionSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run, string var |
+      run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+      succ.asExpr() = run.getScript() and
+      (
+        run.getScript().getAnEnvReachingGitHubEnvWrite(var, _)
+        or
+        run.getScript().getAnEnvReachingGitHubOutputWrite(var, _)
+      )
+    )
+    or
+    exists(Uses step |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = step and
+      succ.asExpr() = step and
+      madSink(succ, "envvar-injection")
+    )
+    or
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate an environment variable. */
+module EnvVarInjectionFlow = TaintTracking::Global<EnvVarInjectionConfig>;

actions/ql/lib/codeql/actions/security/OutputClobberingQuery.qll

@@ -0,0 +1,220 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+private import codeql.actions.security.ArtifactPoisoningQuery
+private import codeql.actions.security.UntrustedCheckoutQuery
+
+abstract class OutputClobberingSink extends DataFlow::Node { }
+
+/**
+ * Holds if a Run step declares a step output variable with contents from a local file.
+ * e.g.
+ *    run: |
+ *      cat test-results/.vars >> $GITHUB_OUTPUT
+ *      echo "sha=$(cat test-results/sha-number)" >> $GITHUB_OUTPUT
+ *      echo "sha=$(<test-results/sha-number)" >> $GITHUB_OUTPUT
+ */
+class OutputClobberingFromFileReadSink extends OutputClobberingSink {
+  OutputClobberingFromFileReadSink() {
+    exists(Run run, Step step, string field1, string field2 |
+      (
+        step instanceof UntrustedArtifactDownloadStep
+        or
+        step instanceof SimplePRHeadCheckoutStep
+      ) and
+      step.getAFollowingStep() = run and
+      this.asExpr() = run.getScript() and
+      // A write to GITHUB_OUTPUT that is not attacker-controlled
+      exists(string str |
+        // The output of a command that is not a file read command
+        run.getScript().getACmdReachingGitHubOutputWrite(str, field1) and
+        not str = run.getScript().getAFileReadCommand()
+        or
+        // A hard-coded string
+        run.getScript().getAWriteToGitHubOutput(field1, str) and
+        str.regexpMatch("[\"'0-9a-zA-Z_\\-]+")
+      ) and
+      // A write to GITHUB_OUTPUT that is attacker-controlled
+      (
+        // echo "sha=$(<test-results/sha-number)" >> $GITHUB_OUTPUT
+        exists(string cmd |
+          run.getScript().getACmdReachingGitHubOutputWrite(cmd, field2) and
+          run.getScript().getAFileReadCommand() = cmd
+        )
+        or
+        // cat test-results/.vars >> $GITHUB_OUTPUT
+        run.getScript().fileToGitHubOutput(_) and
+        field2 = "UNKNOWN"
+      )
+    )
+  }
+}
+
+/**
+ * Holds if a Run step declares an environment variable, uses it in a step variable output.
+ * e.g.
+ *    env:
+ *      BODY: ${{ github.event.comment.body }}
+ *    run: |
+ *      echo "FOO=$BODY" >> $GITHUB_OUTPUT
+ */
+class OutputClobberingFromEnvVarSink extends OutputClobberingSink {
+  OutputClobberingFromEnvVarSink() {
+    exists(Run run, string field1, string field2 |
+      // A write to GITHUB_OUTPUT that is attacker-controlled
+      exists(string var |
+        run.getScript().getAnEnvReachingGitHubOutputWrite(var, field1) and
+        exists(run.getInScopeEnvVarExpr(var)) and
+        run.getScript() = this.asExpr()
+      ) and
+      // A write to GITHUB_OUTPUT that is not attacker-controlled
+      exists(string str |
+        // The output of a command that is not a file read command
+        run.getScript().getACmdReachingGitHubOutputWrite(str, field2) and
+        not str = run.getScript().getAFileReadCommand()
+        or
+        // A hard-coded string
+        run.getScript().getAWriteToGitHubOutput(field2, str) and
+        str.regexpMatch("[\"'0-9a-zA-Z_\\-]+")
+      ) and
+      not field2 = field1
+    )
+  }
+}
+
+/**
+ *      - id: clob1
+ *        env:
+ *          BODY: ${{ github.event.comment.body }}
+ *        run: |
+ *          # VULNERABLE
+ *          echo $BODY
+ *          echo "::set-output name=OUTPUT::SAFE"
+ *      - id: clob2
+ *        env:
+ *          BODY: ${{ github.event.comment.body }}
+ *        run: |
+ *          # VULNERABLE
+ *          echo "::set-output name=OUTPUT::SAFE"
+ *          echo $BODY
+ */
+class WorkflowCommandClobberingFromEnvVarSink extends OutputClobberingSink {
+  string clobbering_var;
+  string clobbered_value;
+
+  WorkflowCommandClobberingFromEnvVarSink() {
+    exists(Run run, string workflow_cmd_stmt, string clobbering_stmt |
+      run.getScript() = this.asExpr() and
+      run.getScript().getAStmt() = clobbering_stmt and
+      clobbering_stmt.regexpMatch("echo\\s+(-e\\s+)?(\"|')?\\$(\\{)?" + clobbering_var + ".*") and
+      exists(run.getInScopeEnvVarExpr(clobbering_var)) and
+      run.getScript().getAStmt() = workflow_cmd_stmt and
+      clobbered_value =
+        trimQuotes(workflow_cmd_stmt.regexpCapture(".*::set-output\\s+name=.*::(.*)", 1))
+    )
+  }
+}
+
+/**
+ *      - id: clob1
+ *        run: |
+ *          # VULNERABLE
+ *          PR="$(<pr-number)"
+ *          echo "$PR"
+ *          echo "::set-output name=OUTPUT::SAFE"
+ *      - id: clob2
+ *        run: |
+ *          # VULNERABLE
+ *          cat pr-number
+ *          echo "::set-output name=OUTPUT::SAFE"
+ *      - id: clob3
+ *        run: |
+ *          # VULNERABLE
+ *          echo "::set-output name=OUTPUT::SAFE"
+ *          ls *.txt
+ *      - id: clob4
+ *        run: |
+ *          # VULNERABLE
+ *          CURRENT_VERSION=$(cat gradle.properties | sed -n '/^version=/ { s/^version=//;p }')
+ *          echo "$CURRENT_VERSION"
+ *          echo "::set-output name=OUTPUT::SAFE"
+ */
+class WorkflowCommandClobberingFromFileReadSink extends OutputClobberingSink {
+  string clobbering_cmd;
+
+  WorkflowCommandClobberingFromFileReadSink() {
+    exists(Run run, string clobbering_stmt |
+      run.getScript() = this.asExpr() and
+      run.getScript().getAStmt() = clobbering_stmt and
+      (
+        // A file's content is assigned to an env var that gets printed to stdout
+        // - run: |
+        //     foo=$(<pr-id.txt)"
+        //     echo "${foo}"
+        exists(string var, string value |
+          run.getScript().getAnAssignment(var, value) and
+          clobbering_cmd = run.getScript().getAFileReadCommand() and
+          trimQuotes(value) = ["$(" + clobbering_cmd + ")", "`" + clobbering_cmd + "`"] and
+          clobbering_stmt.regexpMatch("echo.*\\$(\\{)?" + var + ".*")
+        )
+        or
+        // A file is read and its content is printed to stdout
+        clobbering_cmd = run.getScript().getACommand() and
+        clobbering_cmd.regexpMatch(["ls", Bash::fileReadCommand()] + "\\s.*") and
+        (
+          // - run: echo "foo=$(<pr-id.txt)"
+          clobbering_stmt.regexpMatch("echo.*" + clobbering_cmd + ".*")
+          or
+          // A file content is printed to stdout
+          // - run: cat pr-id.txt
+          clobbering_stmt.indexOf(clobbering_cmd) = 0
+        )
+      )
+    )
+  }
+}
+
+class OutputClobberingFromMaDSink extends OutputClobberingSink {
+  OutputClobberingFromMaDSink() { madSink(this, "output-clobbering") }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate an environment variable.
+ */
+private module OutputClobberingConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source instanceof RemoteFlowSource and
+    not source.(RemoteFlowSource).getSourceType() = "branch"
+  }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof OutputClobberingSink }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run, string var |
+      run.getInScopeEnvVarExpr(var) = pred.asExpr() and
+      succ.asExpr() = run.getScript() and
+      run.getScript().getAWriteToGitHubOutput(_, _)
+    )
+    or
+    exists(Uses step |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = step and
+      succ.asExpr() = step and
+      madSink(succ, "output-clobbering")
+    )
+    or
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      (
+        exists(run.getScript().getAFileReadCommand()) or
+        run.getScript().getAStmt().matches("%::set-output %")
+      )
+    )
+  }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate an environment variable. */
+module OutputClobberingFlow = TaintTracking::Global<OutputClobberingConfig>;

actions/ql/lib/codeql/actions/security/PoisonableSteps.qll

@@ -0,0 +1,56 @@
+import actions
+
+abstract class PoisonableStep extends Step { }
+
+class DangerousActionUsesStep extends PoisonableStep, UsesStep {
+  DangerousActionUsesStep() { poisonableActionsDataModel(this.getCallee()) }
+}
+
+class PoisonableCommandStep extends PoisonableStep, Run {
+  PoisonableCommandStep() {
+    exists(string regexp |
+      poisonableCommandsDataModel(regexp) and
+      this.getScript().getACommand().regexpMatch(regexp)
+    )
+  }
+}
+
+class JavascriptImportUsesStep extends PoisonableStep, UsesStep {
+  JavascriptImportUsesStep() {
+    exists(string script, string line |
+      this.getCallee() = "actions/github-script" and
+      script = this.getArgument("script") and
+      line = script.splitAt("\n").trim() and
+      // const { default: foo } = await import('${{ github.workspace }}/scripts/foo.mjs')
+      // const script = require('${{ github.workspace }}/scripts/test.js');
+      // const script = require('./scripts');
+      line.regexpMatch(".*(import|require)\\(('|\")(\\./|.*github.workspace).*")
+    )
+  }
+}
+
+class SetupNodeUsesStep extends PoisonableStep, UsesStep {
+  SetupNodeUsesStep() {
+    this.getCallee() = "actions/setup-node" and
+    this.getArgument("cache") = "yarn"
+  }
+}
+
+class LocalScriptExecutionRunStep extends PoisonableStep, Run {
+  string path;
+
+  LocalScriptExecutionRunStep() {
+    exists(string cmd, string regexp, int path_group | cmd = this.getScript().getACommand() |
+      poisonableLocalScriptsDataModel(regexp, path_group) and
+      path = cmd.regexpCapture(regexp, path_group)
+    )
+  }
+
+  string getPath() { result = normalizePath(path.splitAt(" ")) }
+}
+
+class LocalActionUsesStep extends PoisonableStep, UsesStep {
+  LocalActionUsesStep() { this.getCallee().matches("./%") }
+
+  string getPath() { result = normalizePath(this.getCallee()) }
+}

actions/ql/lib/codeql/actions/security/RequestForgeryQuery.qll

@@ -0,0 +1,22 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.DataFlow
+
+private class RequestForgerySink extends DataFlow::Node {
+  RequestForgerySink() { madSink(this, "request-forgery") }
+}
+
+/**
+ * A taint-tracking configuration for unsafe user input
+ * that is used to construct and evaluate a system command.
+ */
+private module RequestForgeryConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof RequestForgerySink }
+}
+
+/** Tracks flow of unsafe user input that is used to construct and evaluate a system command. */
+module RequestForgeryFlow = TaintTracking::Global<RequestForgeryConfig>;

actions/ql/lib/codeql/actions/security/SecretExfiltrationQuery.qll

@@ -0,0 +1,21 @@
+private import actions
+private import codeql.actions.TaintTracking
+private import codeql.actions.dataflow.ExternalFlow
+import codeql.actions.dataflow.FlowSources
+import codeql.actions.DataFlow
+
+private class SecretExfiltrationSink extends DataFlow::Node {
+  SecretExfiltrationSink() { madSink(this, "secret-exfiltration") }
+}
+
+/**
+ * A taint-tracking configuration for untrusted data that reaches a sink where it may lead to secret exfiltration
+ */
+private module SecretExfiltrationConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof RemoteFlowSource }
+
+  predicate isSink(DataFlow::Node sink) { sink instanceof SecretExfiltrationSink }
+}
+
+/** Tracks flow of unsafe user input that is used in a context where it may lead to a secret exfiltration. */
+module SecretExfiltrationFlow = TaintTracking::Global<SecretExfiltrationConfig>;

actions/ql/lib/codeql/actions/security/SelfHostedQuery.qll

@@ -0,0 +1,45 @@
+import actions
+
+bindingset[runner]
+predicate isGithubHostedRunner(string runner) {
+  // list of github hosted repos: https://github.com/actions/runner-images/blob/main/README.md#available-images
+  runner
+      .toLowerCase()
+      .regexpMatch("^(ubuntu-([0-9.]+|latest)|macos-([0-9]+|latest)(-x?large)?|windows-([0-9.]+|latest))$")
+}
+
+bindingset[runner]
+predicate is3rdPartyHostedRunner(string runner) {
+  runner.toLowerCase().regexpMatch("^(buildjet|warp)-[a-z0-9-]+$")
+}
+
+/**
+ * This predicate uses data available in the workflow file to identify self-hosted runners.
+ * It does not know if the repository is public or private.
+ * It is a best-effort approach to identify self-hosted runners.
+ */
+predicate staticallyIdentifiedSelfHostedRunner(Job job) {
+  exists(string label |
+    job.getATriggerEvent().getName() =
+      [
+        "issue_comment", "pull_request", "pull_request_review", "pull_request_review_comment",
+        "pull_request_target", "workflow_run"
+      ] and
+    label = job.getARunsOnLabel() and
+    not isGithubHostedRunner(label) and
+    not is3rdPartyHostedRunner(label)
+  )
+}
+
+/**
+ * This predicate uses data available in the job log files to identify self-hosted runners.
+ * It is a best-effort approach to identify self-hosted runners.
+ */
+predicate dynamicallyIdentifiedSelfHostedRunner(Job job) {
+  exists(string runner_info |
+    repositoryDataModel("public", _) and
+    workflowDataModel(job.getEnclosingWorkflow().getLocation().getFile().getRelativePath(), _,
+      job.getId(), _, _, runner_info) and
+    runner_info.indexOf("self-hosted:true") > 0
+  )
+}

actions/ql/lib/codeql/actions/security/UntrustedCheckoutQuery.qll

@@ -0,0 +1,384 @@
+import actions
+private import codeql.actions.DataFlow
+private import codeql.actions.dataflow.FlowSources
+private import codeql.actions.TaintTracking
+
+string checkoutTriggers() {
+  result = ["pull_request_target", "workflow_run", "workflow_call", "issue_comment"]
+}
+
+/**
+ * A taint-tracking configuration for PR HEAD references flowing
+ * into actions/checkout's ref argument.
+ */
+private module ActionsMutableRefCheckoutConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    (
+      // remote flow sources
+      source instanceof GitHubCtxSource
+      or
+      source instanceof GitHubEventCtxSource
+      or
+      source instanceof GitHubEventJsonSource
+      or
+      source instanceof MaDSource
+      or
+      // `ref` argument contains the PR id/number or head ref
+      exists(Expression e |
+        source.asExpr() = e and
+        (
+          containsHeadRef(e.getExpression()) or
+          containsPullRequestNumber(e.getExpression())
+        )
+      )
+      or
+      // 3rd party actions returning the PR head ref
+      exists(StepsExpression e, UsesStep step |
+        source.asExpr() = e and
+        e.getStepId() = step.getId() and
+        (
+          step.getCallee() = "eficode/resolve-pr-refs" and e.getFieldName() = "head_ref"
+          or
+          step.getCallee() = "xt0rted/pull-request-comment-branch" and e.getFieldName() = "head_ref"
+          or
+          step.getCallee() = "alessbell/pull-request-comment-branch" and
+          e.getFieldName() = "head_ref"
+          or
+          step.getCallee() = "gotson/pull-request-comment-branch" and e.getFieldName() = "head_ref"
+          or
+          step.getCallee() = "potiuk/get-workflow-origin" and
+          e.getFieldName() = ["sourceHeadBranch", "pullRequestNumber"]
+          or
+          step.getCallee() = "github/branch-deploy" and e.getFieldName() = ["ref", "fork_ref"]
+        )
+      )
+    )
+  }
+
+  predicate isSink(DataFlow::Node sink) {
+    exists(Uses uses |
+      uses.getCallee() = "actions/checkout" and
+      uses.getArgumentExpr(["ref", "repository"]) = sink.asExpr()
+    )
+  }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+module ActionsMutableRefCheckoutFlow = TaintTracking::Global<ActionsMutableRefCheckoutConfig>;
+
+private module ActionsSHACheckoutConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source.asExpr().getATriggerEvent().getName() =
+      ["pull_request_target", "workflow_run", "workflow_call", "issue_comment"] and
+    (
+      // `ref` argument contains the PR head/merge commit sha
+      exists(Expression e |
+        source.asExpr() = e and
+        containsHeadSHA(e.getExpression())
+      )
+      or
+      // 3rd party actions returning the PR head sha
+      exists(StepsExpression e, UsesStep step |
+        source.asExpr() = e and
+        e.getStepId() = step.getId() and
+        (
+          step.getCallee() = "eficode/resolve-pr-refs" and e.getFieldName() = "head_sha"
+          or
+          step.getCallee() = "xt0rted/pull-request-comment-branch" and e.getFieldName() = "head_sha"
+          or
+          step.getCallee() = "alessbell/pull-request-comment-branch" and
+          e.getFieldName() = "head_sha"
+          or
+          step.getCallee() = "gotson/pull-request-comment-branch" and e.getFieldName() = "head_sha"
+          or
+          step.getCallee() = "potiuk/get-workflow-origin" and
+          e.getFieldName() = ["sourceHeadSha", "mergeCommitSha"]
+        )
+      )
+    )
+  }
+
+  predicate isSink(DataFlow::Node sink) {
+    exists(Uses uses |
+      uses.getCallee() = "actions/checkout" and
+      uses.getArgumentExpr(["ref", "repository"]) = sink.asExpr()
+    )
+  }
+
+  predicate isAdditionalFlowStep(DataFlow::Node pred, DataFlow::Node succ) {
+    exists(Run run |
+      pred instanceof FileSource and
+      pred.asExpr().(Step).getAFollowingStep() = run and
+      succ.asExpr() = run.getScript() and
+      exists(run.getScript().getAFileReadCommand())
+    )
+  }
+}
+
+module ActionsSHACheckoutFlow = TaintTracking::Global<ActionsSHACheckoutConfig>;
+
+bindingset[s]
+predicate containsPullRequestNumber(string s) {
+  exists(
+    normalizeExpr(s)
+        .regexpFind([
+            "\\bgithub\\.event\\.number\\b", "\\bgithub\\.event\\.issue\\.number\\b",
+            "\\bgithub\\.event\\.pull_request\\.id\\b",
+            "\\bgithub\\.event\\.pull_request\\.number\\b",
+            "\\bgithub\\.event\\.check_suite\\.pull_requests\\[\\d+\\]\\.id\\b",
+            "\\bgithub\\.event\\.check_suite\\.pull_requests\\[\\d+\\]\\.number\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.pull_requests\\[\\d+\\]\\.id\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.pull_requests\\[\\d+\\]\\.number\\b",
+            "\\bgithub\\.event\\.check_run\\.pull_requests\\[\\d+\\]\\.id\\b",
+            "\\bgithub\\.event\\.check_run\\.pull_requests\\[\\d+\\]\\.number\\b",
+            // heuristics
+            "\\bpr_number\\b", "\\bpr_id\\b"
+          ], _, _)
+  )
+}
+
+bindingset[s]
+predicate containsHeadSHA(string s) {
+  exists(
+    normalizeExpr(s)
+        .regexpFind([
+            "\\bgithub\\.event\\.pull_request\\.head\\.sha\\b",
+            "\\bgithub\\.event\\.pull_request\\.merge_commit_sha\\b",
+            "\\bgithub\\.event\\.workflow_run\\.head_commit\\.id\\b",
+            "\\bgithub\\.event\\.workflow_run\\.head_sha\\b",
+            "\\bgithub\\.event\\.check_suite\\.after\\b",
+            "\\bgithub\\.event\\.check_suite\\.head_commit\\.id\\b",
+            "\\bgithub\\.event\\.check_suite\\.head_sha\\b",
+            "\\bgithub\\.event\\.check_suite\\.pull_requests\\[\\d+\\]\\.head\\.sha\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.after\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.head_commit\\.id\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.head_sha\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.pull_requests\\[\\d+\\]\\.head\\.sha\\b",
+            "\\bgithub\\.event\\.check_run\\.head_sha\\b",
+            "\\bgithub\\.event\\.check_run\\.pull_requests\\[\\d+\\]\\.head\\.sha\\b",
+            "\\bgithub\\.event\\.merge_group\\.head_sha\\b",
+            "\\bgithub\\.event\\.merge_group\\.head_commit\\.id\\b",
+            // heuristics
+            "\\bhead\\.sha\\b", "\\bhead_sha\\b", "\\bmerge_sha\\b", "\\bpr_head_sha\\b"
+          ], _, _)
+  )
+}
+
+bindingset[s]
+predicate containsHeadRef(string s) {
+  exists(
+    normalizeExpr(s)
+        .regexpFind([
+            "\\bgithub\\.event\\.pull_request\\.head\\.ref\\b", "\\bgithub\\.head_ref\\b",
+            "\\bgithub\\.event\\.workflow_run\\.head_branch\\b",
+            "\\bgithub\\.event\\.check_suite\\.pull_requests\\[\\d+\\]\\.head\\.ref\\b",
+            "\\bgithub\\.event\\.check_run\\.check_suite\\.pull_requests\\[\\d+\\]\\.head\\.ref\\b",
+            "\\bgithub\\.event\\.check_run\\.pull_requests\\[\\d+\\]\\.head\\.ref\\b",
+            "\\bgithub\\.event\\.merge_group\\.head_ref\\b",
+            // heuristics
+            "\\bhead\\.ref\\b", "\\bhead_ref\\b", "\\bmerge_ref\\b", "\\bpr_head_ref\\b",
+            // env vars
+            "GITHUB_HEAD_REF",
+          ], _, _)
+  )
+}
+
+class SimplePRHeadCheckoutStep extends Step {
+  SimplePRHeadCheckoutStep() {
+    // This should be:
+    // artifact instanceof PRHeadCheckoutStep
+    // but PRHeadCheckoutStep uses Taint Tracking anc causes a non-Monolitic Recursion error
+    // so we list all the subclasses of PRHeadCheckoutStep here and use actions/checkout as a workaround
+    // instead of using ActionsMutableRefCheckout and ActionsSHACheckout
+    exists(Uses uses |
+      this = uses and
+      uses.getCallee() = "actions/checkout" and
+      exists(uses.getArgument("ref")) and
+      not uses.getArgument("ref").matches("%base%") and
+      uses.getATriggerEvent().getName() = checkoutTriggers()
+    )
+    or
+    this instanceof GitMutableRefCheckout
+    or
+    this instanceof GitSHACheckout
+    or
+    this instanceof GhMutableRefCheckout
+    or
+    this instanceof GhSHACheckout
+  }
+}
+
+/** Checkout of a Pull Request HEAD */
+abstract class PRHeadCheckoutStep extends Step {
+  abstract string getPath();
+}
+
+/** Checkout of a Pull Request HEAD ref */
+abstract class MutableRefCheckoutStep extends PRHeadCheckoutStep { }
+
+/** Checkout of a Pull Request HEAD ref */
+abstract class SHACheckoutStep extends PRHeadCheckoutStep { }
+
+/** Checkout of a Pull Request HEAD ref using actions/checkout action */
+class ActionsMutableRefCheckout extends MutableRefCheckoutStep instanceof UsesStep {
+  ActionsMutableRefCheckout() {
+    this.getCallee() = "actions/checkout" and
+    (
+      exists(
+        ActionsMutableRefCheckoutFlow::PathNode source, ActionsMutableRefCheckoutFlow::PathNode sink
+      |
+        ActionsMutableRefCheckoutFlow::flowPath(source, sink) and
+        this.getArgumentExpr(["ref", "repository"]) = sink.getNode().asExpr()
+      )
+      or
+      // heuristic base on the step id and field name
+      exists(string value, Expression expr |
+        value.regexpMatch(".*(head|branch|ref).*") and expr = this.getArgumentExpr("ref")
+      |
+        expr.(StepsExpression).getStepId() = value
+        or
+        expr.(SimpleReferenceExpression).getFieldName() = value and
+        not expr instanceof GitHubExpression
+        or
+        expr.(NeedsExpression).getNeededJobId() = value
+        or
+        expr.(JsonReferenceExpression).getAccessPath() = value
+        or
+        expr.(JsonReferenceExpression).getInnerExpression() = value
+      )
+    )
+  }
+
+  override string getPath() {
+    if exists(this.(UsesStep).getArgument("path"))
+    then result = this.(UsesStep).getArgument("path")
+    else result = "GITHUB_WORKSPACE/"
+  }
+}
+
+/** Checkout of a Pull Request HEAD ref using actions/checkout action */
+class ActionsSHACheckout extends SHACheckoutStep instanceof UsesStep {
+  ActionsSHACheckout() {
+    this.getCallee() = "actions/checkout" and
+    (
+      exists(ActionsSHACheckoutFlow::PathNode source, ActionsSHACheckoutFlow::PathNode sink |
+        ActionsSHACheckoutFlow::flowPath(source, sink) and
+        this.getArgumentExpr(["ref", "repository"]) = sink.getNode().asExpr()
+      )
+      or
+      // heuristic base on the step id and field name
+      exists(string value, Expression expr |
+        value.regexpMatch(".*(head|sha|commit).*") and expr = this.getArgumentExpr("ref")
+      |
+        expr.(StepsExpression).getStepId() = value
+        or
+        expr.(SimpleReferenceExpression).getFieldName() = value and
+        not expr instanceof GitHubExpression
+        or
+        expr.(NeedsExpression).getNeededJobId() = value
+        or
+        expr.(JsonReferenceExpression).getAccessPath() = value
+        or
+        expr.(JsonReferenceExpression).getInnerExpression() = value
+      )
+    )
+  }
+
+  override string getPath() {
+    if exists(this.(UsesStep).getArgument("path"))
+    then result = this.(UsesStep).getArgument("path")
+    else result = "GITHUB_WORKSPACE/"
+  }
+}
+
+/** Checkout of a Pull Request HEAD ref using git within a Run step */
+class GitMutableRefCheckout extends MutableRefCheckoutStep instanceof Run {
+  GitMutableRefCheckout() {
+    exists(string cmd | this.getScript().getACommand() = cmd |
+      cmd.regexpMatch("git\\s+(fetch|pull).*") and
+      (
+        (containsHeadRef(cmd) or containsPullRequestNumber(cmd))
+        or
+        exists(string varname, string expr |
+          expr = this.getInScopeEnvVarExpr(varname).getExpression() and
+          (
+            containsHeadRef(expr) or
+            containsPullRequestNumber(expr)
+          ) and
+          exists(cmd.regexpFind(varname, _, _))
+        )
+      )
+    )
+  }
+
+  override string getPath() { result = this.(Run).getWorkingDirectory() }
+}
+
+/** Checkout of a Pull Request HEAD ref using git within a Run step */
+class GitSHACheckout extends SHACheckoutStep instanceof Run {
+  GitSHACheckout() {
+    exists(string cmd | this.getScript().getACommand() = cmd |
+      cmd.regexpMatch("git\\s+(fetch|pull).*") and
+      (
+        containsHeadSHA(cmd)
+        or
+        exists(string varname, string expr |
+          expr = this.getInScopeEnvVarExpr(varname).getExpression() and
+          containsHeadSHA(expr) and
+          exists(cmd.regexpFind(varname, _, _))
+        )
+      )
+    )
+  }
+
+  override string getPath() { result = this.(Run).getWorkingDirectory() }
+}
+
+/** Checkout of a Pull Request HEAD ref using gh within a Run step */
+class GhMutableRefCheckout extends MutableRefCheckoutStep instanceof Run {
+  GhMutableRefCheckout() {
+    exists(string cmd | this.getScript().getACommand() = cmd |
+      cmd.regexpMatch(".*(gh|hub)\\s+pr\\s+checkout.*") and
+      (
+        (containsHeadRef(cmd) or containsPullRequestNumber(cmd))
+        or
+        exists(string varname |
+          (
+            containsHeadRef(this.getInScopeEnvVarExpr(varname).getExpression()) or
+            containsPullRequestNumber(this.getInScopeEnvVarExpr(varname).getExpression())
+          ) and
+          exists(cmd.regexpFind(varname, _, _))
+        )
+      )
+    )
+  }
+
+  override string getPath() { result = this.(Run).getWorkingDirectory() }
+}
+
+/** Checkout of a Pull Request HEAD ref using gh within a Run step */
+class GhSHACheckout extends SHACheckoutStep instanceof Run {
+  GhSHACheckout() {
+    exists(string cmd | this.getScript().getACommand() = cmd |
+      cmd.regexpMatch("gh\\s+pr\\s+checkout.*") and
+      (
+        containsHeadSHA(cmd)
+        or
+        exists(string varname |
+          containsHeadSHA(this.getInScopeEnvVarExpr(varname).getExpression()) and
+          exists(cmd.regexpFind(varname, _, _))
+        )
+      )
+    )
+  }
+
+  override string getPath() { result = this.(Run).getWorkingDirectory() }
+}

actions/ql/lib/codeql/actions/security/UseOfKnownVulnerableActionQuery.qll

@@ -0,0 +1,22 @@
+import actions
+
+class KnownVulnerableAction extends UsesStep {
+  string vulnerable_action;
+  string fixed_version;
+  string vulnerable_version;
+  string vulnerable_sha;
+
+  KnownVulnerableAction() {
+    vulnerableActionsDataModel(vulnerable_action, vulnerable_version, vulnerable_sha, fixed_version) and
+    this.getCallee() = vulnerable_action and
+    (this.getVersion() = vulnerable_version or this.getVersion() = vulnerable_sha)
+  }
+
+  string getFixedVersion() { result = fixed_version }
+
+  string getVulnerableAction() { result = vulnerable_action }
+
+  string getVulnerableVersion() { result = vulnerable_version }
+
+  string getVulnerableSha() { result = vulnerable_sha }
+}

actions/ql/lib/codeql/actions/security/UseOfUnversionedImmutableAction.qll

@@ -0,0 +1,28 @@
+import actions
+
+class UnversionedImmutableAction extends UsesStep {
+  string immutable_action;
+
+  UnversionedImmutableAction() {
+    isImmutableAction(this, immutable_action) and
+    not isSemVer(this.getVersion())
+  }
+}
+
+bindingset[version]
+predicate isSemVer(string version) {
+  // https://semver.org/#is-there-a-suggested-regular-expression-regex-to-check-a-semver-string with optional v prefix
+  version
+      .regexpMatch("^v?(0|[1-9]\\d*)\\.(0|[1-9]\\d*)\\.(0|[1-9]\\d*)(?:-((?:0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*)(?:\\.(?:0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*))*))?(?:\\+([0-9a-zA-Z-]+(?:\\.[0-9a-zA-Z-]+)*))?$") or
+  // or N or N.x or N.N.x with optional v prefix
+  version.regexpMatch("^v?[1-9]\\d*$") or
+  version.regexpMatch("^v?[1-9]\\d*\\.(x|0|([1-9]\\d*))$") or
+  version.regexpMatch("^v?[1-9]\\d*\\.(0|([1-9]\\d*))\\.(x|0|([1-9]\\d*))$") or
+  // or latest which will work
+  version = "latest"
+}
+
+predicate isImmutableAction(UsesStep actionStep, string actionName) {
+  immutableActionsDataModel(actionName) and
+  actionStep.getCallee() = actionName
+}

actions/ql/lib/codeql/files/FileSystem.qll

@@ -0,0 +1,177 @@
+/** Provides classes for working with files and folders. */
+
+private import codeql.Locations
+
+/** A file or folder. */
+abstract class Container extends @container {
+  /** Gets a file or sub-folder in this container. */
+  Container getAChildContainer() { this = result.getParentContainer() }
+
+  /** Gets a file in this container. */
+  File getAFile() { result = this.getAChildContainer() }
+
+  /** Gets a sub-folder in this container. */
+  Folder getAFolder() { result = this.getAChildContainer() }
+
+  /**
+   * Gets the absolute, canonical path of this container, using forward slashes
+   * as path separator.
+   *
+   * The path starts with a _root prefix_ followed by zero or more _path
+   * segments_ separated by forward slashes.
+   *
+   * The root prefix is of one of the following forms:
+   *
+   *   1. A single forward slash `/` (Unix-style)
+   *   2. An upper-case drive letter followed by a colon and a forward slash,
+   *      such as `C:/` (Windows-style)
+   *   3. Two forward slashes, a computer name, and then another forward slash,
+   *      such as `//FileServer/` (UNC-style)
+   *
+   * Path segments are never empty (that is, absolute paths never contain two
+   * contiguous slashes, except as part of a UNC-style root prefix). Also, path
+   * segments never contain forward slashes, and no path segment is of the
+   * form `.` (one dot) or `..` (two dots).
+   *
+   * Note that an absolute path never ends with a forward slash, except if it is
+   * a bare root prefix, that is, the path has no path segments. A container
+   * whose absolute path has no segments is always a `Folder`, not a `File`.
+   */
+  abstract string getAbsolutePath();
+
+  /**
+   * Gets the base name of this container including extension, that is, the last
+   * segment of its absolute path, or the empty string if it has no segments.
+   *
+   * Here are some examples of absolute paths and the corresponding base names
+   * (surrounded with quotes to avoid ambiguity):
+   *
+   * <table border="1">
+   * <tr><th>Absolute path</th><th>Base name</th></tr>
+   * <tr><td>"/tmp/tst.go"</td><td>"tst.go"</td></tr>
+   * <tr><td>"C:/Program Files (x86)"</td><td>"Program Files (x86)"</td></tr>
+   * <tr><td>"/"</td><td>""</td></tr>
+   * <tr><td>"C:/"</td><td>""</td></tr>
+   * <tr><td>"D:/"</td><td>""</td></tr>
+   * <tr><td>"//FileServer/"</td><td>""</td></tr>
+   * </table>
+   */
+  string getBaseName() {
+    result = this.getAbsolutePath().regexpCapture(".*/(([^/]*?)(?:\\.([^.]*))?)", 1)
+  }
+
+  /**
+   * Gets the extension of this container, that is, the suffix of its base name
+   * after the last dot character, if any.
+   *
+   * In particular,
+   *
+   *  - if the name does not include a dot, there is no extension, so this
+   *    predicate has no result;
+   *  - if the name ends in a dot, the extension is the empty string;
+   *  - if the name contains multiple dots, the extension follows the last dot.
+   *
+   * Here are some examples of absolute paths and the corresponding extensions
+   * (surrounded with quotes to avoid ambiguity):
+   *
+   * <table border="1">
+   * <tr><th>Absolute path</th><th>Extension</th></tr>
+   * <tr><td>"/tmp/tst.go"</td><td>"go"</td></tr>
+   * <tr><td>"/tmp/.classpath"</td><td>"classpath"</td></tr>
+   * <tr><td>"/bin/bash"</td><td>not defined</td></tr>
+   * <tr><td>"/tmp/tst2."</td><td>""</td></tr>
+   * <tr><td>"/tmp/x.tar.gz"</td><td>"gz"</td></tr>
+   * </table>
+   */
+  string getExtension() {
+    result = this.getAbsolutePath().regexpCapture(".*/([^/]*?)(\\.([^.]*))?", 3)
+  }
+
+  /** Gets the file in this container that has the given `baseName`, if any. */
+  File getFile(string baseName) {
+    result = this.getAFile() and
+    result.getBaseName() = baseName
+  }
+
+  /** Gets the sub-folder in this container that has the given `baseName`, if any. */
+  Folder getFolder(string baseName) {
+    result = this.getAFolder() and
+    result.getBaseName() = baseName
+  }
+
+  /** Gets the parent container of this file or folder, if any. */
+  Container getParentContainer() { containerparent(result, this) }
+
+  /**
+   * Gets the relative path of this file or folder from the root folder of the
+   * analyzed source location. The relative path of the root folder itself is
+   * the empty string.
+   *
+   * This has no result if the container is outside the source root, that is,
+   * if the root folder is not a reflexive, transitive parent of this container.
+   */
+  string getRelativePath() {
+    exists(string absPath, string pref |
+      absPath = this.getAbsolutePath() and sourceLocationPrefix(pref)
+    |
+      absPath = pref and result = ""
+      or
+      absPath = pref.regexpReplaceAll("/$", "") + "/" + result and
+      not result.matches("/%")
+    )
+  }
+
+  /**
+   * Gets the stem of this container, that is, the prefix of its base name up to
+   * (but not including) the last dot character if there is one, or the entire
+   * base name if there is not.
+   *
+   * Here are some examples of absolute paths and the corresponding stems
+   * (surrounded with quotes to avoid ambiguity):
+   *
+   * <table border="1">
+   * <tr><th>Absolute path</th><th>Stem</th></tr>
+   * <tr><td>"/tmp/tst.go"</td><td>"tst"</td></tr>
+   * <tr><td>"/tmp/.classpath"</td><td>""</td></tr>
+   * <tr><td>"/bin/bash"</td><td>"bash"</td></tr>
+   * <tr><td>"/tmp/tst2."</td><td>"tst2"</td></tr>
+   * <tr><td>"/tmp/x.tar.gz"</td><td>"x.tar"</td></tr>
+   * </table>
+   */
+  string getStem() {
+    result = this.getAbsolutePath().regexpCapture(".*/([^/]*?)(?:\\.([^.]*))?", 1)
+  }
+
+  /**
+   * Gets a URL representing the location of this container.
+   *
+   * For more information see https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/#providing-urls.
+   */
+  abstract string getURL();
+
+  /**
+   * Gets a textual representation of the path of this container.
+   *
+   * This is the absolute path of the container.
+   */
+  string toString() { result = this.getAbsolutePath() }
+}
+
+/** A folder. */
+class Folder extends Container, @folder {
+  override string getAbsolutePath() { folders(this, result) }
+
+  /** Gets the URL of this folder. */
+  override string getURL() { result = "folder://" + this.getAbsolutePath() }
+}
+
+/** A file. */
+class File extends Container, @file {
+  override string getAbsolutePath() { files(this, result) }
+
+  /** Gets the URL of this file. */
+  override string getURL() { result = "file://" + this.getAbsolutePath() + ":0:0:0:0" }
+
+  /** Holds if this file was extracted from ordinary source code. */
+  predicate fromSource() { any() }
+}

actions/ql/lib/ext/config/argument_injection_sinks.yml

@@ -0,0 +1,15 @@
+extensions:
+  - addsTo:
+      pack: codeql/actions-all
+      extensible: argumentInjectionSinksDataModel
+    # https://gtfobins.github.io/
+    # https://0xn3va.gitbook.io/cheat-sheets/web-application/command-injection/argument-injection
+    data:
+      - ["(awk)\\s(.*?)", 1, 2]
+      - ["(find)\\s(.*?)", 1, 2]
+      - ["(git clone)\\s(.*?)", 1, 2]
+      - ["(sed)\\s(.*?)", 1, 2]
+      - ["(tar)\\s(.*?)", 1, 2]
+      - ["(wget)\\s(.*?)", 1, 2]
+      - ["(zip)\\s(.*?)", 1, 2]
+

actions/ql/lib/ext/config/context_event_map.yml

@@ -0,0 +1,53 @@
+extensions:
+  - addsTo:
+      pack: codeql/actions-all
+      extensible: contextTriggerDataModel 
+    data:
+      - ["commit_comment", "github.event.comment"] 
+      - ["commit_comment", "github.event.changes"] 
+      - ["discussion", "github.event.discussion"]
+      - ["discussion", "github.event.changes"] 
+      - ["discussion_comment", "github.event.comment"] 
+      - ["discussion_comment", "github.event.discussion"]
+      - ["discussion_comment", "github.event.changes"] 
+      - ["issues", "github.event.issue"] 
+      - ["issues", "github.event.changes"] 
+      - ["issue_comment", "github.event.issue"] 
+      - ["issue_comment", "github.event.comment"] 
+      - ["issue_comment", "github.event.changes"] 
+      - ["gollum", "github.event.pages"] 
+      - ["gollum", "github.event.changes"] 
+      - ["pull_request_comment", "github.event.comment"] 
+      - ["pull_request_comment", "github.event.pull_request"] 
+      - ["pull_request_comment", "github.head_ref"] 
+      - ["pull_request_comment", "github.event.changes"] 
+      - ["pull_request_review", "github.event.pull_request"] 
+      - ["pull_request_review", "github.event.review"] 
+      - ["pull_request_review", "github.head_ref"] 
+      - ["pull_request_review", "github.event.changes"] 
+      - ["pull_request_review_comment", "github.event.comment"] 
+      - ["pull_request_review_comment", "github.event.pull_request"] 
+      - ["pull_request_review_comment", "github.event.review"] 
+      - ["pull_request_review_comment", "github.head_ref"] 
+      - ["pull_request_review_comment", "github.event.changes"] 
+      - ["pull_request_target", "github.event.pull_request"] 
+      - ["pull_request_target", "github.head_ref"] 
+      - ["pull_request_target", "github.event.changes"] 
+      - ["push", "github.event.commits"]
+      - ["push", "github.event.head_commit"]
+      - ["push", "github.event.changes"] 
+      - ["workflow_run", "github.event.workflow"]
+      - ["workflow_run", "github.event.workflow_run"]
+      - ["workflow_run", "github.event.changes"] 
+      # workflow_call receives the same event payload as the calling workflow
+      - ["workflow_call", "github.event.comment"] 
+      - ["workflow_call", "github.event.discussion"]
+      - ["workflow_call", "github.event.inputs"]
+      - ["workflow_call", "github.event.issue"] 
+      - ["workflow_call", "github.event.pages"] 
+      - ["workflow_call", "github.event.pull_request"] 
+      - ["workflow_call", "github.event.review"] 
+      - ["workflow_call", "github.event.workflow"]
+      - ["workflow_call", "github.event.workflow_run"]
+      - ["workflow_call", "github.event.changes"] 
+