Menu
Full autopilot — plan, implement, and polish a task end-to-end.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
| name | cook |
| description | Full autopilot — plan, implement, and polish a task end-to-end. |
| allowed-tools | Bash, Read, Glob, Grep, AskUserQuestion, Agent |
End-to-end delivery. Scale the process to the task.
Triage the task first (see workflow rules), then follow this pipeline. Scale each step to the complexity — a simple task needs a one-line plan, a complex one needs a full design document.
/plan. It will ask the user clarifying questions via AskUserQuestion before producing a plan — this is mandatory, never skip it even for simple tasks. Scale the plan's depth to the complexity./research to fill them, then revise the plan and confirm again./execute with the plan file path. It handles parallelism, subagents, and failure recovery./commit to stage and commit the changes./polish. It reviews and amends iteratively until 3x LGTM. If it reveals design gaps that require rethinking (not just code fixes), loop back to step 1. On completion, it produces a report. Skip for simple tasks./pr.Amend the last commit with current changes and/or an updated message.
Stage changes and create a commit with a conventional commit message.
Fetch and display diffs for various scopes (PR, commit, branch, staged, working tree). Used by other skills that need to analyze changes.
Execute a plan file — parse phases, spawn parallel subagents, handle failures.
Start or end a forked session for exploring rabbit holes without losing the main conversation's context.
Structured design intake for non-trivial tasks. Asks clarifying questions and produces a design proposal.