Superpowers
A complete software development methodology for your coding agents - built on composable skills that trigger automatically.
Skills
Each skill triggers automatically at the right moment. Read in order, or jump to what you need.
Using Superpowers
The bootstrap: how skills are found and triggered. Invoke relevant skills before any response.
Brainstorming
Turn ideas into designs through dialogue. No code until a design is presented and approved.
Writing Plans
Write bite-sized implementation plans for an engineer with zero context. DRY, YAGNI, TDD.
Executing Plans
Load, critically review, and execute a written plan with review checkpoints.
Subagent-Driven Development
The core loop: fresh subagent per task, task review, broad final review.
Test-Driven Development
Write the test first. Watch it fail. Write minimal code to pass.
Systematic Debugging
Always find root cause before fixing. Symptom fixes are failure.
Requesting Code Review
Dispatch a reviewer subagent with precisely crafted context. Review early, review often.
Receiving Code Review
Verify before implementing. Technical correctness over social comfort.
Verification Before Completion
Evidence before claims, always. No completion claims without fresh verification.
Finishing a Development Branch
Verify tests, detect environment, present options, execute choice, clean up.
Dispatching Parallel Agents
One agent per independent problem domain. Let them work concurrently.
Using Git Worktrees
Ensure work happens in an isolated workspace. Detect first, then native tools, then git.
Writing Skills
TDD applied to process documentation. Write test cases, watch them fail, write the skill.