| name | playbook |
| description | mokata · Run the full v1 story end-to-end on this repo (integration check). |
| when_to_use | Engage when the user EXPLICITLY asks to run mokata's full pipeline end-to-end as an integration check, when they want to smoke-test the whole brainstorm-to-ship story on a repo, or when validating a fresh install performs a complete run. Do NOT engage implicitly or on your own initiative — it is a long, multi-write flow — and do NOT run it on a dirty working tree without confirmation. |
mokata Agent Skill. This is mokata's playbook capability, surfaced so Claude can engage it
automatically when the moment fits. It runs the SAME protocol as the /mokata:playbook command,
from one shared source — follow that protocol directly here; do not hand off to a parallel
flow. mokata's non-negotiables still hold: durable writes are human-gated (preview, then
explicit approval), and this capability's own gate is never silently skipped.
⛭ mokata playbook active — gate: a long multi-write flow — run on explicit request only; confirm on a dirty tree
mokata · playbook (the whole story, end-to-end)
Drive mokata's full v1 pipeline end-to-end on this repo as an integration check — every
phase and gate in sequence. --parallel runs it with subagents (degrading to sequential when
the harness has none); --dense compresses sub-agent hand-backs (F4 output density). The
gates and audit ledger apply throughout.
How to run
Resolve the bundled engine (read ~/.mokata/plugin-root → ROOT, or a mokata CLI on PATH):
PY="$(command -v python3 || command -v python)"
ENGINE="PYTHONPATH=\"$ROOT/src\" \"$PY\" -m mokata"
eval "$ENGINE playbook --path ."
eval "$ENGINE playbook --parallel --path ."
eval "$ENGINE playbook --parallel --fanout --dense --path ."
It announces the active stage (banner) at the start and on completion, and reports each
phase's result. Watch a parallel run live with /mokata:progress --lanes or /mokata:watch.
Rationalizations — stop if you catch yourself thinking any of these
| Excuse | Reality |
|---|
| "This seems like a good moment to run the whole flow." | Run only on an EXPLICIT request — never implicitly, it's a long multi-write flow. |
| "The tree's a bit dirty but it'll be fine." | Confirm on a dirty tree before running — uncommitted work is at risk. |
| "It's just an integration check — writes don't need gating." | Every durable write it performs is still human-gated. |
Verification — confirm each before you claim this skill is done
Evidence, not "seems right" — check every box or say which is unmet and why:
Contract
CAN
- run the full v1 story end-to-end as an integration check
MUST NOT
- run implicitly — explicit user request only; or run on a dirty tree without confirmation (advisory)
- skip the human gate on any durable write it performs (gate: write-gate)
DEPENDS ON
- the full toolchain and a repo the user accepts test writes in (advisory)
Grounding: (gate: …) boundaries are enforced by that gate in code; (advisory) ones are protocol discipline this skill follows, not a hard block.