apk-tool

Use at the start of a manual executor session to find and claim a ready bead. Do not invoke in planner sessions or inside swarm workers.

Use after implementation and verification are complete to close the bead, create follow-up beads, and commit the updated `.beads/` state. This is the final step of a manual executor session.

Break a discussed or approved problem into Beads epics and tasks with clear dependencies and validation work. Use when the user wants to turn a problem statement, planning discussion, or approved execution plan into a Beads structure instead of ad-hoc TODOs.

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements, and design before implementation.

Use after implementing changes to run the exact verification commands from the current execution plan. This stage-1 version is intentionally generic and should be specialized per repo once the runtime workflow is clear.

Execute one assigned bead inside a swarm coordinated by swarm-epic. Use only when a coordinator has assigned the bead, file scope, and verification contract; the worker implements, verifies, and reports without mutating bead state.

Run repeated manual executor cycles scoped to a single epic: pick the next ready descendant bead under that epic, execute it, then continue until the epic has no ready descendants or a blocker requires user input. Use when the user wants sequential epic progress without swarm coordination.

Run repeated manual executor cycles bead-by-bead until the ready queue is exhausted or a blocker requires user input. Use when the user wants sequential autonomous progress without swarm coordination.

Run exactly one full executor cycle for one bead: claim, write a local execution plan, implement, verify, and close. Use when the user wants to execute a single bead end-to-end.

Use after all work on a feature branch is complete and verified. Pushes the branch and creates a PR targeting main.

Run a planner-only Beads session: brainstorm, produce or confirm an execution plan, get user approval, then create beads and stop. Use when the user wants to turn a current problem or topic into Beads without implementing.

Planner-only research step for resolving factual unknowns after brainstorming and before Beads creation. Use only when unresolved technical or domain uncertainty would materially weaken the plan.

Run an epic-level review after swarm or sequential epic execution. Use after all intended child beads are complete to classify remaining issues as P1, P2, or P3 and decide whether the epic can close or needs follow-up beads.

Coordinate multi-agent execution for one epic. Use after planning when the user wants parallel work across ready descendant beads with coordinator-owned Beads state, automatic branch setup, automatic epic review, runtime files, and Agent Mail reservations.

Validate a planned epic before autonomous execution. Use after beads-planner and before swarm-epic to check dependency quality, bead size, file scope, verification instructions, and parallel-safety notes.

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

Use when you have a spec or requirements for a multi-step task, before touching code. This is an executor skill - use only when a bead has been claimed and you are in an executor session, NOT in a planner session.