Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Added usage tutorial: step-by-step guide, strategy table, parameter reference, token cost estimates, and resume support docs.
What This Skill Does
Cross-references GitHub PRs and issues to surface duplicate PRs and issues that already have a solving PR but no explicit link. Spawns parallel Sonnet subagents to analyze up to 1000 PRs and 1000 issues simultaneously, then groups findings into thematic clusters scored by actionability. After review, it can comment, label, or close items via the GitHub CLI.
Manual cross-referencing stops being feasible past a few dozen items; parallel subagent analysis covers 1000 PRs and 1000 issues in minutes instead of hours.
When to use it
- Finding duplicate bug-fix PRs submitted by different contributors
- Linking open issues to merged PRs that solve them but lack a fixes #N reference
- Auditing cross-references before closing out a milestone
- Cleaning up a backlog of stale duplicate PRs in a large repo
- Checking which open issues already have a PR in flight
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Find duplicate PRs and missing issue links in myorg/myrepo
- 1Fetch PR and issue metadata from GitHub API using fetch-data.sh, producing compact per-line indexes and an existing-references map
- 2Split PRs into batches of 50 and spawn parallel Sonnet subagents, each receiving its batch plus the full issue and PR indexes
- 3Merge subagent results, deduplicate symmetric duplicate-PR pairs, filter by confidence threshold, and group into thematic clusters scored by actionability
- 4Spawn a single verification subagent to confirm candidates using full PR diffs and issue bodies fetched via gh CLI, downgrading or removing false positives
- 5Present a scored cluster report and prompt the user to choose comment, label, or close actions per cluster before anything is executed
A markdown report listing clusters of related findings sorted by actionability score, with suggested actions per cluster and a separate section for items flagged for manual review
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.