Releases: maruwork/github-optimization
v1.2.12
v1.2.11
GitHub hosted hardening now aligns with the shelf's own release expectations: secret scanning, push protection, Dependabot security updates, and master branch protection are enabled. actions/checkout references are updated to v6, stale distribution examples and verdict wording are corrected, and collector regressions were rerun on June 19, 2026.
v1.2.10
Branch version metadata now reflects current master. Git Bash hosted metadata and latest CI evidence now match the PowerShell collector's compact schema. PowerShell and Git Bash regulation regressions plus hosted dogfood verification were rerun on June 19, 2026.
v1.2.9
- Stabilize Windows hosted-evidence regression assertions across ABSENT, PASS, and API_BLOCKED issue-template cases.
- Escape literal gate-text checks correctly so GitHub Actions Windows CI matches local verification.
- Re-run PowerShell and Git Bash regulation regressions before cutting the release.
v1.2.8
- Fix the Windows hosted-evidence regression test so it no longer depends on JSON property order.
- Keep release metadata aligned with the current shelf HEAD after the follow-up CI fix.
- Re-run PowerShell and Git Bash regression coverage for the updated release state.
v1.2.7
v1.2.6
v1.2.6
Formal release for the GitHub Optimization audit shelf.
Changes
- Fixed the Gitleaks evidence artifact boundary so environment execution denials are reported as SKIPPED instead of BLOCKED where direct scoring evidence is required.
- Added regression coverage for PowerShell and Git Bash collector behavior around Gitleaks access-denied artifacts.
- Completed strict all-file self-audit for the shelf state at this release.
Verification
- validate-regulation-index.ps1: PASS
- check-tracked-files.ps1: PASS
- check-gitignore-consistency.ps1: PASS
- run-audit-quickstart.ps1: PASS
- PowerShell regulation tests: PASS
- Git Bash regulation tests: PASS
- GitHub CI / CodeQL: PASS
Notes
This release keeps raw collector evidence separate from scoring evidence. Managed sandbox or Git Bash execution artifacts must not be treated as final G-01 scoring when a direct Gitleaks transcript is required.