oss-issue-matcherlisted
Install: claude install-skill NarenDawar/narens-claude-toolkit
# OSS Issue Matcher
## Overview
Most "good first issue" lists rank by label, not by whether anyone will answer your pull request. This skill scores each candidate repo on how it has actually treated outside PRs, then picks issues only from repos that pass. Every number comes from `scripts/repo_pulse.py`, never from memory. Everything is read-only: do not comment on issues, claim them, fork or open PRs.
## Setup
Run `gh auth status`. If gh is missing or not logged in, stop and tell the user to run `! gh auth login`.
Take the user's language or stack, skills, hours per week and goal (learn, résumé, fix a tool they use). If no language or skill was given, ask that one question; otherwise state your defaults in one line and proceed.
## Process
1. **Find candidates.** Run two searches, `--label "good first issue"` and `--label "help wanted"`:
`gh search issues --label "good first issue" --state open --no-assignee --archived=false --language <lang> --sort updated --limit 100 --json repository,number,title,updatedAt --jq '.[] | [.repository.nameWithOwner, .number, .title, .updatedAt] | @tsv'`
Count the issues per repo. Keep up to 12 repos, ranked by how many matching issues they have (a maintainer who keeps labelling issues is active), then by the most recent update. Prefer repos with 2 or more issues; a repo with one auto-generated-looking issue is usually too new to score. Add repos the user names.
2. **Score the repos.** Run `python scripts/repo_pulse.py owner/a owne