What problem does it solve? Incoming GitHub issues pile up without consistent classification, specification, or routing, forcing maintainers to manually evaluate every bug report and feature request before work can begin. ## Core Features & Use Cases - Label-Based State Machine: Moves issues through defined states (needs-triage, needs-info, ready-for-agent, ready-for-human, wontfix) with enforced transition rules and conflict detection. - Interactive Grilling Sessions: Interviews the maintainer to build complete specifications, resuming from prior triage notes without re-asking resolved questions. - Agent Brief Generation: Writes durable, behavior-focused agent briefs so issues marked ready-for-agent can be picked up by an AFK agent weeks later. - Out-of-Scope Knowledge Base: Records rejected feature requests in .out-of-scope/ files and matches new issues against prior rejections to avoid re-litigating decisions. - Use Case: A maintainer asks "show me anything that needs my attention" and receives a grouped summary of unlabeled, needs-triage, and needs-info issues, then triages issue #42 through reproduction, grilling, and a ready-for-agent brief. ## Quick Start Ask the agent to show all GitHub issues that need attention in the current repository and triage them one by one.