crowi-pr-received

Triages external GitHub pull requests via remote diffs with supply-chain attack safeguards.

1.1k|165|Updated Aug 18, 2014
One-click install
npx skills add https://github.com/crowi/crowi --skill crowi-pr-received
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: crowi-pr-received
Source: https://github.com/crowi/crowi/tree/main/.claude/skills/crowi-pr-received
Command: npx skills add https://github.com/crowi/crowi --skill crowi-pr-received

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

External pull requests to a maintainer-driven project arrive unpredictably and carry real risk: spam, lockfile poisoning, malicious workflow changes, and fake security fixes. This Skill standardizes how a maintainer triages, classifies, and responds to outside PRs without ever checking out untrusted code.

Core Features & Use Cases

  • Remote-only triage: Classifies PRs (spam, deps/lockfile, workflows, docs, i18n, bugfix, feature, security) using only gh pr view and gh pr diff, with cross-repository author activity checks to detect spam signals.
  • Supply-chain defense rules: Enforces hard rules such as never merging workflow changes, treating any pnpm-lock.yaml modification with maximum suspicion, and forbidding script execution until a full diff review completes.
  • Approval-gated responses and isolated integration: Drafts English reply templates for maintainer approval before any comment/close/merge, and routes accepted code through an isolated worktree with --ignore-scripts installs and authorship-preserving merge discipline.
  • Use Case: A new locale translation PR arrives from an unknown contributor. The Skill checks the author's cross-repo activity, classifies it as i18n, and drafts a conditional-accept response asking about real usage and ongoing maintenance commitment before any code is touched.

Quick Start

Ask the assistant to triage pull request number 914 on the crowi repository using this skill and produce a classification with a draft response.

Frequently Asked Questions about crowi-pr-received

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I safely triage an external GitHub pull request?

Use gh pr view and gh pr diff to review the remote diff without checking out the contributor's branch. Classify the PR into categories like spam, deps, docs, bugfix, or feature, then draft a response for maintainer approval before any outbound action.

How to detect spam pull requests on GitHub?

Query the author's cross-repository activity with the GitHub search API for recent PRs. Signals include identical PR titles spread across many repositories, no prior issue or discussion, and an account whose only activity is distributing such PRs.

Why should you never run pnpm install on an unreviewed pull request?

Install scripts, test runners, and build configs all execute repository code, so running them on unreviewed changes can execute malicious payloads. Lockfile changes can also redirect resolved package URLs to hostile tarballs, a technique called lockfile poisoning.

Should maintainers accept pull requests that modify GitHub Actions workflows?

No. Workflow changes are a classic path for stealing CI secrets, so they should never be merged from external contributors regardless of category. If the underlying idea has value, maintainers should reimplement the workflow themselves.

When should a new translation locale pull request be accepted?

Accept only when there is evidence of real user demand and a contributor committed to maintaining the locale long-term. A locale without an active maintainer decays as new strings are added, so a conditional-accept response with a roughly two-week response window is the default.

How should a public pull request claiming a security fix be handled?

Treat it as public disclosure and redirect the reporter to private vulnerability reporting such as GitHub Security Advisories, then close the PR. Implement the fix internally, credit the reporter, and scrutinize the diff carefully since fake security fixes can hide backdoors.