make_plan

Create a detailed, multi-document implementation plan for a feature or task before any code is written.

What it does

make_plan drives a mandatory clarifying-questions interview, enforces a hard Zero-Ambiguity Gate, and produces a plans/<feature>/ document set ending in a task-by-task execution plan.

The document set typically includes an ambiguity register (the gate's audit trail), an index, requirements, a current-state analysis, one or more component technical specs, a testing strategy with concrete specification test cases, and a 99-execution-plan.md that structures the work into phases and sessions following specification-first test ordering.

Every decision in the plan traces back to an explicit, user-confirmed entry in the ambiguity register — nothing is guessed on your behalf.

When to use it

• You want to plan a feature, refactor, or task before building it.
• You want a spec/plan written down with clear scope and acceptance criteria.
• Not for executing an existing plan — use exec_plan for that.

Trigger phrases

Auto-triggers on: "make a plan", "make_plan", "plan this feature", "create an implementation plan", "plan out this work", "write a spec/plan for X". Explicit: /codeops:make_plan.

Worked example

/codeops:make_plan

Describe a small feature when prompted. The skill runs the clarifying interview, presents the ambiguity register for your confirmation (the gate), and — only once you confirm — writes plans/<feature>/. To implement it:

/codeops:exec_plan <feature-name>

Related skills

• grill_me — deep disambiguation before planning.
• make_requirements — produces RDs that make_plan can consume.
• preflight — audit the plan before you build.
• exec_plan — execute the finished plan.
• upgrade_plan — bring an older plan up to current standards.