ci(hypatia-scan): emit + upload SARIF so scans reconcile code-scanning alerts#485
Merged
Conversation
…g alerts The reusable scan produced hypatia-findings.json as an *artifact* only and declared `security-events: read`, so it could neither create nor CLOSE code-scanning alerts. Alerts uploaded by an earlier SARIF path (tool "Hypatia", category "hypatia") were therefore orphaned: once the underlying code was fixed, the fix never propagated to the Security tab and the alerts lingered as stale "open" findings (observed on metadatastician/gossamer — ~40 of 108 open alerts were already fixed in code but could not auto-close). This restores the designed behaviour that lib/hypatia/sarif.ex documents: - run the scan a second time with HYPATIA_FORMAT=sarif -> hypatia.sarif - upload it via github/codeql-action/upload-sarif with category "hypatia" - grant security-events: write SARIF render already filters the code_scanning_alerts meta-rules (lib/hypatia/sarif.ex @meta_rule_modules), so re-uploading cannot self-echo (the 7,724-alert loop the render-time filter was added to close). The upload step is `continue-on-error: true`, so callers that still grant only security-events: read are unaffected — and because every caller SHA-pins this reusable, the change is opt-in: it takes effect for a repo only when that repo bumps its pin AND grants write. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
hyperpolymath
added a commit
to metadatastician/gossamer
that referenced
this pull request
Jul 11, 2026
…rite) (#111) ## What Makes gossamer's Hypatia scan **upload SARIF** to code scanning, so code-scanning alerts **self-reconcile**. - reusable pin `d7c2271` → **`2ccc38e`** (standards `main`, post-[standards#485](hyperpolymath/standards#485)) - `permissions: security-events` **read → write** - `secrets: inherit` (lets the reusable use `HYPATIA_SCAN_PAT` for Dependabot reads; falls back to `GITHUB_TOKEN`) ## Why The scan previously uploaded `hypatia-findings.json` as an **artifact only**, so it could neither create nor **close** code-scanning alerts. Findings fixed in code stayed "open" forever — that's how **~40 already-fixed alerts** (timeouts, `believe_me`, SD007/SD022, workflow singletons, scorecard checks) got stuck among the 108 open alerts. With this, every scan uploads SARIF (tool `Hypatia`, `category: hypatia`, matching the historical `analysis_key`), and GitHub auto-closes anything fixed since the last run. No self-echo — the SARIF renderer filters the `code_scanning_alerts` meta-rules. ## Note The one-off backlog of 108 alerts was already **reconciled manually** (dismissed with per-alert reasons: stale→`won't fix`+remediation note, inherent-FFI→`won't fix`, `.envrc`→`false positive`). This PR makes the pipeline **self-healing** so it doesn't recur. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The reusable Hypatia scan now emits SARIF and uploads it to code scanning (
category: hypatia), and declaressecurity-events: write.Why
The reusable scan produced
hypatia-findings.jsonas an artifact only and declaredsecurity-events: read. So it could neither create nor close code-scanning alerts. Alerts that an earlier SARIF path had uploaded (toolHypatia, categoryhypatia) were orphaned: once the underlying code was fixed, nothing propagated the fix to the Security tab, and the alerts lingered as stale "open" findings.Surfaced on metadatastician/gossamer: of 108 open alerts, ~40 were already fixed in code (timeouts,
believe_me, SD007/SD022, all four workflow singletons, scorecard StaticAnalysis/DependencyPinning) but could not auto-close. The rest were inherent FFI-boundary casts + derivative meta-findings.How
Restores the behaviour that
lib/hypatia/sarif.exalready documents ("the workflow can call this format directly"):HYPATIA_FORMAT=sarif→hypatia.sarif.github/codeql-action/upload-sarifwithcategory: hypatia(matches the historicalanalysis_key, so GitHub reconciles — fixed findings auto-close).permissions: security-events: write.Safety
lib/hypatia/sarif.exfilters thecode_scanning_alertsmeta-rules (@meta_rule_modules) at render time — the exact loop the render-time filter was added to close (the 7,724-alert incident). Uploading is now safe.security-events: write.continue-on-error: true; a caller still granting onlysecurity-events: readskips it gracefully.Follow-up
metadatastician/gossamer will bump its
hypatia-scan.ymlpin to this reusable's merged SHA and grantsecurity-events: write, so its scans self-reconcile going forward (the 108 alerts were dismissed manually as an immediate fix).🤖 Generated with Claude Code