test-driven-development
The RED-GREEN-REFACTOR spine of the methodology (pattern-test-driven-development): write one minimal failing test → watch it fail for the right reason → write the simplest code to pass → watch it pass with pristine output → refactor while green → repeat.
The Iron Law: “NO PRODUCTION CODE WITHOUT A FAILING TEST FIRST.” Wrote code before the test? “Delete it. Start over.” — no keeping it “as reference”, no “adapting” it while writing the test. “If you didn’t watch the test fail, you don’t know if it tests the right thing.”
Its length is almost entirely pattern-anti-rationalization: an exhaustive excuse→reality table (“too simple to test”, “I’ll test after”, “already manually tested”, “deleting X hours is wasteful” = sunk-cost fallacy, “TDD is dogmatic, I’m being pragmatic”) plus a Red-Flags “STOP and start over” list. It carries a companion testing-anti-patterns.md (testing mocks instead of real behavior, test-only methods on production classes), a “when stuck” table (hard-to-test = hard-to-use design signal), and a verification checklist. Each green step ends in a commit → artifact-atomic-commit.
Superpowers is the fourth+ framework in the TDD cluster, alongside mp-tdd (red→green, refactor moved to review), addy-tdd (red-green-refactor + Prove-It), plus GSD’s --tdd flag and Spec Kit’s constitution-mandated TDD — all clustered at pattern-test-driven-development. Superpowers’ is among the strictest, matching Spec Kit’s non-negotiability but enforcing it through rhetoric (delete- and-restart) rather than a constitution gate.
See Also
- sp-writing-plans — bakes this red-green micro-loop into every task’s steps.
- sp-subagent-driven-development · sp-executing-plans — the executors that run this per task.
- sp-systematic-debugging — its Phase-4 fix step writes a failing test via this skill.
- mp-tdd · addy-tdd — the cross-framework TDD cluster.
- pattern-test-driven-development · pattern-anti-rationalization — the techniques applied.
- stage-implement — the canonical stage this implements.