Install
$ agentstack add skill-graysurf-agent-kit-fix-bug-pr ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Fix Bug PR
Contract
Prereqs:
- Run inside the target git repo.
gitandghavailable onPATH, andgh auth statussucceeds.python3available onPATH(helper scripts parse/patch PR Markdown deterministically).- Start from a clean working tree (avoid staging unrelated local changes).
Inputs:
- Optional PR number to operate on; otherwise resolve an open bug-type PR automatically.
- Optional label filters to narrow candidate PRs (defaults:
bug,type: bug).
Outputs:
- If a matching PR exists:
- Code fix committed and pushed to the PR head branch.
- PR comment describing what changed, tests run, and which bug IDs were updated.
- PR body updated so the Issues Found table reflects latest bug Status; if all bugs are fixed, set overall
Status: fixed. - If no bug-type PR exists (or none with unresolved bugs): no repo changes; report “no related issues”.
Exit codes:
- N/A (workflow skill). Helper scripts define their own exit codes.
Failure modes:
- No matching bug-type PR found, or PR body does not include a parseable
## Issues Foundtable. - Missing tooling (
gh/git/python3) orghnot authenticated. - Dirty working tree, merge conflicts, or failing tests prevent a safe commit/push.
- High-risk areas (auth/billing/migrations/deploy) should stop or be skipped.
Test-First Evidence Gate
- Before editing production behavior, capture failing-test evidence or an explicit waiver.
- This gate applies before editing production behavior.
- For bug items, prefer a focused failing regression, unit, integration, or acceptance test that demonstrates the unresolved issue.
- Record the command, exit code, failing test file/name, and concise failure summary before production edits.
- If no failing test is practical, record why, what substitute validation will run, and the residual risk.
- PR comments/body updates must include
Change classification,Failing test before fix,Final validation, andWaiver reasonwhen
applicable.
Workflow
- Resolve the target PR:
- Auto-pick an open PR with unresolved bug items:
$AGENT_HOME/skills/automation/bug/fix-bug-pr/scripts/bug-pr-resolve.sh- Or target a specific PR:
$AGENT_HOME/skills/automation/bug/fix-bug-pr/scripts/bug-pr-resolve.sh --pr- If it exits
2: stop and report “No relevant issues found (no bug-type PR found).”
- Checkout the PR branch:
gh pr checkout
- Identify the next unresolved bug item:
- Use
next_bug_idfrombug-pr-resolve.shoutput. - Fix one bug item per run unless multiple items share the same root cause.
- Implement the minimal fix + validation:
- Capture the Test-First Evidence Gate before production edits.
- Prefer small, targeted diffs; avoid refactors.
- Follow the target repo’s testing/build docs and CI workflow: install required tooling/deps, then run the most relevant lint/test/build
commands.
- If docs are missing/unclear, infer likely commands from common entrypoints (package.json scripts, Makefile/Justfile/Taskfile, language
defaults).
- Treat validation as a gate: if checks fail, do not commit/push; fix and re-run (or stop and report). If checks cannot be run, document
why in the PR comment.
- Commit + push:
- Use
semantic-commit-autostage(end-to-end automation) unless the user explicitly wants manual staging. - Push to the PR head branch.
- Update PR body progress (status sync):
- Mark the fixed bug ID(s) as
fixedand recompute overall status: $AGENT_HOME/skills/automation/bug/fix-bug-pr/scripts/bug-pr-patch.sh --pr --mark-fixed- If all bug items are fixed, the PR body will be updated to
Status: fixed.
- Comment on the PR with what changed:
- Include: fixed bug IDs, summary, tests run, and any remaining open items.
- Use
skills/automation/bug/fix-bug-pr/references/COMMENT_TEMPLATE.mdas the structure.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: graysurf
- Source: graysurf/agent-kit
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.