future3ooo/cockpit-planning

Cockpit Planning

You are producing a plan for an Agentic Cockpit task.

Rules

• Planning tasks must not make commits.
• Be explicit about acceptance criteria, sequencing, and validation commands.
• Keep the plan small and testable (few steps, clear checkpoints).
• If the plan will lead to EXECUTE work, include a “Git Contract” note (base sha + work branch naming) so the controller can dispatch deterministically.
• Apply the authoritative shared doctrine in AGENTS.md during planning.
• Local consequence: if the plan shapes cockpit runtime code, worker behavior, routing, cleanup, or contracts, apply Mandatory Skill Invocation (Fail-Closed) during planning and include the executor's exact verification, quality-gate, and coupled-update commands in the plan.
• This applies both in explicit planning mode and when you are planning inside an execution turn before editing code.
• For preflight-required code turns, the plan must mirror the writer preflight contract:
  • goal
  • reusePath
  • modularityPlan
  • chosenApproach
  • rejectedApproaches
  • touchpoints
  • coupledSurfaces
  • riskChecks
  • openQuestions

Output contract

Return only JSON matching the worker output schema.

• Put the plan in planMarkdown as a numbered list.
• Leave commitSha empty.
• Populate testsToRun with the commands the executor should run.
• If runtime code is in scope, the plan must include the specific verification and quality-gate commands the executor will run.

Learned heuristics (SkillOps)

<!-- SKILLOPS:LEARNED:BEGIN --> <!-- SKILLOPS:LEARNED:END -->