You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Two failure patterns drove the 21.4% failure rate:
Actionlint download instability (primary β ~7 failures): The old ci.yml downloaded actionlint via curl, which hit transient GitHub CDN 504 errors (exit code 22). This was already fixed in the current ci.yml by switching to gh release download with the comment "Uses gh release download rather than direct CDN curl to avoid transient 504s from GitHub's release asset CDN."
Gitleaks false positives (ongoing β ~5 failures): The secret-scan job with fetch-depth: 0 causes gitleaks to scan all git history in PR context (600-1000+ commits). Branches that included commits from the old frameworks/ git subtree (since removed via chore: remove frameworks directory (#101)) triggered the generic-api-key rule on test fixture files:
frameworks/bmad-method/test/adversarial-review-tests/sample-content.md (line 24) β sample content with "token": "value"
frameworks/gsd/tests/commands.test.cjs (line 759) β test fixture
Scope
Add .gitleaks.toml to suppress false positives from the frameworks/ git-subtree directory (removed from tree but present in history), and add a BATS test to validate the config structure.
Implementation Checklist
Create .gitleaks.toml with path-based allowlist for frameworks/
Create tests/test_gitleaks_config.bats to validate config presence and structure
Add tests/test_gitleaks_config.bats to .github/workflows/lint.yml BATS test list
Tests to Write
tests/test_gitleaks_config.bats: assert .gitleaks.toml exists, parses as valid TOML, and includes frameworks/ in the allowlist paths
Two failure patterns drove the 21.4% failure rate:
Actionlint download instability (~7 failures, already fixed): The old ci.yml downloaded actionlint via curl, hitting transient GitHub CDN 504 errors (exit code 22). The current ci.yml already contains the fix β gh release download with the explicit comment "Uses gh release download rather than direct CDN curl to avoid transient 504s from GitHub's release asset CDN."
Gitleaks false positives (~5 failures, fixed by this PR): The secret-scan job scans PR history (600β1000+ commits) and finds fake API keys in test fixture files from the old frameworks/ git subtree (removed in commit d42bfe7). Affected files: frameworks/bmad-method/test/adversarial-review-tests/sample-content.md, frameworks/gsd/sdk/src/query/config-query.test.ts, frameworks/gsd/sdk/src/query/config-mutation.test.ts, frameworks/gsd/tests/commands.test.cjs.
Plan Execution
Analyzed 200 historical CI runs to identify failure patterns
Confirmed false positives are in frameworks/ test fixtures (not real credentials)
Wrote failing BATS test (tests/test_gitleaks_config.bats) β red phase confirmed
Created .gitleaks.toml with frameworks/ path allowlist β green phase confirmed
Added tests/test_gitleaks_config.bats to lint.yml BATS test list
bash scripts/dev-lead-lint.sh β all checks passed
yamllint on all workflow files β passed
Test Results
[lint] shellcheck on 40 file(s)...
[lint] validating 4 agent profile(s)...
OK: agents/agentic-workflows.md
OK: agents/compliance-auditor.md
OK: agents/feature-ideator.md
OK: agents/pr-reviewer.md
[lint] all checks passed
yamllint: exit 0 (no errors)
Python test validation:
Test 1: .gitleaks.toml exists at repo root β PASS
Test 2: .gitleaks.toml is valid TOML β PASS
Test 3: .gitleaks.toml allowlist covers frameworks/ path β PASS
Files Changed
.gitleaks.toml (new): Extends default gitleaks config with a global path allowlist for frameworks/ β suppresses false positives from git-subtree test fixtures in historical commits
.github/workflows/lint.yml: Added tests/test_gitleaks_config.bats to the BATS test list
Notes
The [extend] useDefault = true directive preserves the full default gitleaks ruleset β only the frameworks/ path is exempted, not any rule globally. Push events to main are unaffected (gitleaks only scans the latest commit on push). The allowlist protects PR branches that include the historical frameworks/ subtree commits from surfacing false positives going forward.
Activity
feat(fleet-monitor): track high-failure workflows as GitHub Issues
don-petry commentedon May 21, 2026
dev-lead - Investigate and fix
feat(fleet-monitor): track high-failure workflows as GitHub Issues
feat(fleet-monitor): track high-failure workflows as GitHub Issues
feat(fleet-monitor): track high-failure workflows as GitHub Issues
don-petry commentedon Jun 9, 2026
don-petry commentedon Jun 9, 2026
Dev-Lead Implementation Plan
Issue: #244 β [Fleet Monitor] petry-projects/.github-private β ci.yml
Root Cause Analysis
Two failure patterns drove the 21.4% failure rate:
Actionlint download instability (primary β ~7 failures): The old
ci.ymldownloaded actionlint viacurl, which hit transient GitHub CDN 504 errors (exit code 22). This was already fixed in the currentci.ymlby switching togh release downloadwith the comment "Uses gh release download rather than direct CDN curl to avoid transient 504s from GitHub's release asset CDN."Gitleaks false positives (ongoing β ~5 failures): The
secret-scanjob withfetch-depth: 0causes gitleaks to scan all git history in PR context (600-1000+ commits). Branches that included commits from the oldframeworks/git subtree (since removed viachore: remove frameworks directory (#101)) triggered thegeneric-api-keyrule on test fixture files:frameworks/bmad-method/test/adversarial-review-tests/sample-content.md(line 24) β sample content with"token": "value"frameworks/gsd/sdk/src/query/config-query.test.ts(line 228) βconst apiKey = '...'frameworks/gsd/sdk/src/query/config-mutation.test.ts(line 501) βconst apiKey = '...'frameworks/gsd/tests/commands.test.cjs(line 759) β test fixtureScope
Add
.gitleaks.tomlto suppress false positives from theframeworks/git-subtree directory (removed from tree but present in history), and add a BATS test to validate the config structure.Implementation Checklist
.gitleaks.tomlwith path-based allowlist forframeworks/tests/test_gitleaks_config.batsto validate config presence and structuretests/test_gitleaks_config.batsto.github/workflows/lint.ymlBATS test listTests to Write
tests/test_gitleaks_config.bats: assert.gitleaks.tomlexists, parses as valid TOML, and includesframeworks/in the allowlist pathsTest Command
bash scripts/dev-lead-lint.shdon-petry commentedon Jun 9, 2026
Dev-Lead: Implementation Complete
Root Cause Summary
Two failure patterns drove the 21.4% failure rate:
Actionlint download instability (~7 failures, already fixed): The old
ci.ymldownloaded actionlint viacurl, hitting transient GitHub CDN 504 errors (exit code 22). The currentci.ymlalready contains the fix βgh release downloadwith the explicit comment "Uses gh release download rather than direct CDN curl to avoid transient 504s from GitHub's release asset CDN."Gitleaks false positives (~5 failures, fixed by this PR): The
secret-scanjob scans PR history (600β1000+ commits) and finds fake API keys in test fixture files from the oldframeworks/git subtree (removed in commitd42bfe7). Affected files:frameworks/bmad-method/test/adversarial-review-tests/sample-content.md,frameworks/gsd/sdk/src/query/config-query.test.ts,frameworks/gsd/sdk/src/query/config-mutation.test.ts,frameworks/gsd/tests/commands.test.cjs.Plan Execution
frameworks/test fixtures (not real credentials)tests/test_gitleaks_config.bats) β red phase confirmed.gitleaks.tomlwithframeworks/path allowlist β green phase confirmedtests/test_gitleaks_config.batstolint.ymlBATS test listbash scripts/dev-lead-lint.shβ all checks passedyamllinton all workflow files β passedTest Results
Files Changed
.gitleaks.toml(new): Extends default gitleaks config with a global path allowlist forframeworks/β suppresses false positives from git-subtree test fixtures in historical commitstests/test_gitleaks_config.bats(new): BATS tests validating config presence, TOML validity, and allowlist coverage.github/workflows/lint.yml: Addedtests/test_gitleaks_config.batsto the BATS test listNotes
The
[extend] useDefault = truedirective preserves the full default gitleaks ruleset β only theframeworks/path is exempted, not any rule globally. Push events tomainare unaffected (gitleaks only scans the latest commit on push). The allowlist protects PR branches that include the historicalframeworks/subtree commits from surfacing false positives going forward.34 remaining items
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦
feat: implement issue #244 β [Fleet Monitor] petry-projects/.github-pβ¦