AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Manage Prs

skill-billchirico-bills-agent-skills-manage-prs · by BillChirico

Manage and maintain GitHub pull requests end-to-end. Use when asked to fix PR review comments, unresolved review threads, top-level PR conversation comments, requested changes, merge conflicts, or CI/CD failures; when the user says to manage, maintain, clean up, unblock, get green, finish, or prepare a PR; or when a PR must have no unresolved actionable comments and no failing checks.

No reviews yet
0 installs
22 views
0.0% view→install

Install

$ agentstack add skill-billchirico-bills-agent-skills-manage-prs

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No 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.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-billchirico-bills-agent-skills-manage-prs)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

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 →
Are you the author of Manage Prs? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Manage PRs

Operating Rule

Treat GitHub as the source of truth. Do not rely on stale browser state, old summaries, local guesses, or a previous agent's report.

Only mark review threads resolved after:

  1. The related code/doc/test fix is committed.
  2. The branch is pushed.
  3. The live PR state is re-read and the fixed thread IDs are still the threads being resolved.

Top-level PR conversation comments do not have review-thread resolution. Address them by fixing the issue, pushing the fix, and replying only when a reply is useful or explicitly requested.

Quick Start

From the target repo:

python3 "/scripts/inspect_pr_state.py" --repo "." --pr ""

If the current branch has exactly one PR, omit --pr.

After fixes are pushed and the live thread IDs are confirmed:

python3 "/scripts/resolve_review_threads.py" --repo "." --thread-id ""

Workflow

  1. Confirm the target PR.
  • Use a provided PR number/URL when present.
  • Otherwise run gh pr view --json number,url,headRefName,headRepositoryOwner.
  • Run git status --short before editing. Preserve unrelated user changes.
  1. Refresh from GitHub.
  • Run git fetch --all --prune.
  • Inspect mergeability with gh pr view --json number,url,headRefName,baseRefName,mergeable,mergeStateStatus,reviewDecision.
  • If the branch is behind or conflicted, update from the PR base using the repo's normal merge/rebase pattern before calling the PR clean.
  1. Inventory every blocker.
  • Run inspect_pr_state.py.
  • Read unresolved review threads.
  • Read top-level PR conversation comments.
  • Read review summary bodies, especially CHANGES_REQUESTED.
  • Read failing and pending checks.
  • For failing GitHub Actions checks, fetch logs with gh run view --log or use the existing gh-fix-ci skill/script if available.
  • For external CI providers, open the details URL if tooling is available; otherwise report the exact external URL and continue with local reproduction where possible.
  1. Cluster comments by root cause.
  • Multiple comments often point at one bug. Fix the bug once instead of making one-off edits for each comment.
  • Separate actionable requests from stale, duplicate, or already-addressed comments.
  • Do not resolve a stale-looking thread until live code proves it is obsolete and the branch has been pushed.
  1. Fix with tests.
  • Reproduce failing CI locally when feasible.
  • Add or update focused tests for review-comment fixes that affect behavior.
  • Run the narrowest meaningful verification first, then broader gates when the PR touches shared code.
  • Do not weaken lint, typecheck, tests, snapshots, coverage, security checks, or required workflows to get green.
  1. Commit and push.
  • Use the repo's commit conventions.
  • If push is rejected, fetch/rebase or merge the remote branch, rerun the relevant checks, then push again.
  • Never resolve review threads before this push succeeds.
  1. Re-check GitHub.
  • Re-run inspect_pr_state.py.
  • Re-run gh pr checks or the provider-specific status command.
  • If checks are pending, wait or watch them when practical. Do not claim there are no failing checks while required checks are still red.
  1. Resolve completed review threads.
  • Resolve only the exact thread IDs whose requested changes were fixed and pushed.
  • Use resolve_review_threads.py or the GraphQL mutation directly.
  • Re-run inspect_pr_state.py and confirm unresolved actionable review threads are zero.
  1. Final report.
  • Include the PR URL/number.
  • List the comments fixed, top-level comments addressed, and CI failures cleared.
  • State the exact verification run.
  • If anything remains pending or externally blocked, say exactly which check/comment and link the details.

Comment Handling Rules

  • Review threads: fix, push, then resolve via GraphQL resolveReviewThread.
  • Top-level PR conversation comments: fix the underlying issue. Reply only when a human needs a note; do not pretend they can be resolved.
  • Review summary bodies: treat CHANGES_REQUESTED as actionable even when there is no inline thread.
  • Bot comments: handle if they identify a real blocker. Ignore noise only after proving it is non-actionable.
  • Duplicates: mention they are duplicates in the final report, but still verify there is no separate unresolved request.

CI/CD Handling Rules

  • Required checks must not be failing when done.
  • A local green test does not replace a red GitHub check. Use logs to connect the local fix to the failing workflow.
  • Treat pending checks as pending, not green.
  • Treat skipped/neutral checks as their own bucket. They are not failures unless branch protection or the PR UI marks them as blockers.
  • If a failure is environmental or third-party, gather the run URL, error text, and retry status before reporting it as blocked.
  • If several PRs share the same failing workflow, fix the shared workflow once and verify each affected PR state separately.

Useful GraphQL Mutation

Use this only after the fix is pushed:

gh api graphql \
  -f threadId="" \
  -f query='mutation($threadId:ID!){resolveReviewThread(input:{threadId:$threadId}){thread{id isResolved}}}'

Bundled Scripts

  • scripts/inspect_pr_state.py: prints unresolved review threads, top-level PR comments, review summaries, and status checks for a PR.
  • scripts/resolve_review_threads.py: resolves explicit review-thread IDs after the branch has been fixed and pushed.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.