What problem does it solve? Teams running Fabrik (a GitHub-Project-driven SDLC orchestrator that dispatches Claude Code workers through Specify/Research/Plan/Implement/Review/Validate stages) need a way to supervise the pipeline without reading engine internals. This Skill gives Claude ambient awareness of the board, labels, and stage semantics so it can answer PM-style questions and advise on interventions. ## Core Features & Use Cases - Board and issue inspection: Bundled bash scripts (board.sh, issue.sh, comments.sh, stage-output.sh) query the GitHub Project board, issue state, comments, and stage output via the gh CLI, auto-locating .fabrik/config.yaml. - Label-based control guidance: Explains engine-managed labels (fabrik:locked, stage:X:failed) versus user-set control labels (fabrik:yolo, fabrik:cruise, fabrik:paused, model:opus, base:<branch>) so users steer the pipeline correctly. - Issue authoring and dependency wiring: Coaches users to write worker-actionable issue specs and safely wire blocked_by dependencies via the GitHub API before moving issues into Specify. - Use Case: A user asks "why is #42 stuck?" — Claude runs scripts/issue.sh 42, sees stage:plan:failed, reads the worker output via stage-output.sh, and recommends applying fabrik:extend-turns or commenting to redirect. ## Quick Start Ask Claude what's currently on the Fabrik board or why a specific issue number is stuck, and it will inspect the project state and explain what to do next.