What problem does it solve? Manually reviewing every open issue and pull request in a busy GitHub repository is slow and inconsistent. This Skill automates read-only triage by classifying each item and dispatching one background executor agent per item to produce evidence-backed reports. ## Core Features & Use Cases - Parallel per-item analysis: Spawns one background executor agent per issue or PR, batched at up to 5 concurrent agents, with run-size bounding and a SKIPPED.md overflow log. - Classification and verdicts: Classifies items as question, bug, feature, or other, and assesses PR merge safety against six explicit conditions including CI status, review decision, and conflicts. - Mandatory evidence citations: Every factual code claim must cite a GitHub permalink containing a commit SHA, otherwise it is marked UNVERIFIED. - Use Case: A maintainer returning to a repository with 80 open issues and 20 open PRs runs the triage to get per-item reports and a SUMMARY.md highlighting confirmed bugs, merge-safe PRs, and items needing human decisions. ## Quick Start Ask the agent to triage the open issues and pull requests in the current GitHub repository and summarize which items need attention.