go-finish — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited go-finish (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Guide completion of a feature after the full pipeline has run. Verify everything, clean up, and integrate.
Before presenting options, run the full verification suite:
Read the Verification Commands pattern in patterns.md when running this.
If anything fails: Stop. Do not proceed. Report the failure and suggest whether go-fixer or go-debugger is appropriate.
If skipped tests remain: These are unimplemented test stubs from scaffolding. Either they were intentionally deferred (user decision) or they were missed. Ask the user.
Read .plan/<feature-slug>/FEATURE.md and check every item in "Definition of Done":
Read the Acceptance Criteria Check pattern in patterns.md when writing this.
For each item: either point to evidence (test name, command output) or flag it as unverified.
Sequential safety: This step modifies .claude/skills/doc-project/SKILL.md, which is a singleton file. If another feature's go-finish is running concurrently, you WILL create merge conflicts. This is safe because go-product-manager drives features sequentially — but if you were invoked manually outside that flow, verify no other feature is in progress before updating.
Read .claude/skills/doc-project/SKILL.md. Update:
Then read .claude/skills/doc-project/contexts/<context>.md for each context touched by this feature. Update:
If the context doc doesn't exist, create it at .claude/skills/doc-project/contexts/<context>.md following the context doc template in patterns.md.
Keep updates factual — list what exists, don't interpret. The doc is a map, not a narrative.
Read all .plan/<feature-slug>/task-*_SUMMARY.md files and .plan/<feature-slug>/task-rev-* (reviewer task files) to produce a concise report:
Read the Review Summary Report pattern in patterns.md when writing this.
Read all .plan/<feature-slug>/task-*_SUMMARY.md files. For each task, extract:
Write .feedback/<feature-slug>/feedback.md following the Feedback Template.
This is factual compilation, not interpretation. List what happened, not what should change. The retrospective agent handles interpretation.
Ask: "Should I archive the .plan/<feature-slug>/ directory?"
.plan/done/<feature-slug>/ — preserves history.plan/<feature-slug>/ entirelyRead the Integration Options pattern in patterns.md when presenting this.
#### Option 1: Merge Locally Read the Merge Locally pattern in patterns.md when executing this.
#### Option 2: Push and Create PR Read the Push and Create PR pattern in patterns.md when executing this.
#### Option 3: Keep As-Is Report: "Branch preserved. Plan artifacts at .plan/<feature-slug>/."
#### Option 4: Discard Require typed confirmation. Read the Discard Confirmation pattern in patterns.md when presenting this.
.claude/skills/adr-*/) should NOT be cleaned up. They persist as guidance for future features.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.