Stats
Actions
Tags
From completely
Overview and entry point for the completely harness — a quality-first agent workflow unifying GSD (planning depth), Ralph (autonomous loop), and Beads (the spine), under deterministic gate hooks and a default-FAIL evaluator. Use to see what's installed, the command surface, and how to start.
How this skill is triggered — by the user, by Claude, or both
Slash command
/completely:completelyThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**completely** makes it hard for an agent to quietly cut a corner, fake "done", or ship a stub.
completely makes it hard for an agent to quietly cut a corner, fake "done", or ship a stub.
Short CLI: cmp. Slash: /completely:*.
/completely:init — scaffold the thin layer (DoD, CLAUDE.md rules, project gate) into this repo./completely:sync (cmp sync) — migrate markdown task state into Beads, idempotently./completely:run — drive bd ready: supervised (GSD wave subagents) or unattended (Ralph loop)./completely:doctor (cmp doctor) — upstream version drift + overlay health.Beads = the spine (status + memory via comments/notes/remember + coordination via swarm/gate/merge-slot).
GSD = planning depth (discuss→plan→plan-checker loop). Ralph = the OS-level autonomous loop.
completely = the quality floor (gate hooks + default-FAIL evaluator + no-stub contract) under all of them.
GSD and Ralph are two autonomy modes of the same loop over bd ready, not a choice of tools.
See plugin/core/HARNESS.md (the principle), plugin/core/roles.md (who owns what),
and docs/TOOL-COMPATIBILITY.md (the full design + per-tool detail).
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub 23ag1/completely --plugin completely