superpowers-plan_skill

Overview

This skill writes a precise implementation plan for non-trivial code changes, breaking work into small, verifiable steps and listing exact files to modify. It enforces incremental deliverables, verification commands, and a clear rollback strategy so changes are safe and auditable. Use it before making multi-file edits or any change that affects behavior, data, auth, billing, or production workflows.

How this skill works

Given a change request, the skill outputs a structured plan using the exact format: Goal, Assumptions, Plan (ordered numbered steps), Risks & mitigations, and Rollback plan. Each step lists the files to touch, the minimal change bullets, and precise verification commands or checks. Steps are short (2–10 minutes) and always include verification and rollback notes where applicable.

When to use it

• Any multi-file code change or refactor.
• Changes that impact behavior, data models, authentication, billing, or production flows.
• Debugging that requires systematic isolation and reproducible checks.
• When you need an auditable, incremental rollout with clear verification.

Best practices

• Keep each step small and independently verifiable (2–10 minutes each).
• Prefer incremental deliverables; avoid big-bang edits.
• Always include exact files and exact verification commands.
• Call out assumptions up front and list rollback actions early.
• Include risk severity and a simple mitigation for each risk.

Example use cases

• Add a new API endpoint that touches multiple modules and requires auth and schema changes.
• Refactor a shared utility used across services with targeted tests and runtime checks.
• Modify a data migration that updates production data and needs verification queries.
• Introduce a billing flag that affects metering logic and reporting pipelines.