What problem does it solve? When a task finishes, deciding what to work on next requires re-reading scattered project docs and half-remembered priorities. This Skill reads the repository's documented candidates and operational state to produce a read-only, evidence-grounded recommendation without ever selecting, defining, or implementing work itself. ## Core Features & Use Cases - Three advisory modes: a bare candidate list, a fast ranked suggestion from loaded documentation (default), and a verified suggestion that checks candidates against current repository reality. - Candidate-list hygiene: flags stale, superseded, contradicted, or partially done candidates and surfaces documented-but-untracked work as capture suggestions. - Readiness ranking with lenses: orders candidates by observable readiness, separates waiting-on-other-work from blocked items, and supports quick-win or cleanup ranking lenses. - Use Case: After finishing a task, ask what to work on next and receive a ranked ready list with a recommendation, tie-break rationale, and the pickup command (do or define) for the suggested candidate. ## Quick Start Ask the assistant what to work on next to get a ranked recommendation from the documented candidates.
