2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill repo-pr-triage- _meta.json280 B
- SKILL.md4.1 KB
Overview
This skill scores and prioritizes open GitHub pull requests and issues against a project-specific vision document. It supports onboarding (interviewing the repo owner to build a vision and rubric), scanning (rule-based scoring and duplicate detection), and reporting (actionable Markdown triage reports). It works with any repository via the gh CLI and outputs JSON and Markdown you can act on or automate.
How this skill works
Onboard collects repo context and an owner interview to produce a vision.md and a customized rubric. Scan fetches open PRs with gh, applies a base-50 heuristic rubric with positive/negative modifiers, and detects likely duplicates, producing detailed JSON with scores and reasoning. Report generates categorized Markdown files (prioritize, review, close, summary) so maintainers can act quickly or feed results into automation.
When to use it
- When you need to prioritize a backlog of open PRs against project goals
- Before release or milestone to decide what to fast-track
- To batch-process and clean up stale or low-value PRs
- When onboarding new maintainers to capture project priorities
- To run automated weekly triage via cron or CI
Best practices
- Run onboarding once per repo and update the vision after major shifts in goals
- Adjust rubric weights in the vision/rubric files to reflect project values and risk tolerance
- Review PRs scoring 40–60 manually; ambiguous cases benefit from additional human or LLM review
- Limit scan count to a reasonable number for large repos and run incremental scans regularly
- Use generated Markdown reports as the source for automated notifications or triage meetings
Example use cases
- Set up weekly cron to scan a repo, generate reports, and post the summary to team chat
- Onboard a new maintainer by capturing the project’s mission and auto-prioritizing incoming contributions
- Audit open PRs before a release to identify security or critical bug fixes to fast-track
- De-duplicate similar PRs and surface active contributors for recognition or mentoring
- Run a one-off cleanup to identify PRs likely to close or require rework
FAQ
A GitHub repo URL, an onboarding vision.md produced for that repo, and gh CLI authentication.
How are scores computed?
A base score of 50 is adjusted with rubric modifiers (e.g., security +20, spam -30) using heuristic checks like keyword matching, diff size, and test mentions.
Can I customize the rubric or thresholds?
Yes. Update the vision/rubric files produced during onboarding to change weights and threshold categories.