Superpowers for Spec-First AI Coding
Superpowers enforces spec-first discipline for AI coding agents through brainstorming, specs, plans, TDD, verification, and repeatable review evidence.
Paginated browsing for the full Spec Coding article archive. These pages keep long lists readable while preserving direct links.
Superpowers enforces spec-first discipline for AI coding agents through brainstorming, specs, plans, TDD, verification, and repeatable review evidence.
Define database schemas before writing migrations, including columns, constraints, indexes, API alignment, and rollout order.
AI coding tools drift without constraints, adding fields, renaming functions, expanding scope, and inventing tests. Spec-first prompts keep changes reviewable.
Govern AI-assisted coding with spec-driven prompts: define scope, boundaries, evidence, and audit trails before generated code reaches review.
Review AI-generated pull requests against acceptance criteria: inspect the diff, run evidence checks, and catch failures a quick skim misses.
Use a pre-merge risk register for AI-generated code: flag auth, data, contract, migration, rollback, and observability risks.
Use test-evidence gates for AI-generated code: require meaningful tests before merge and catch hallucinated implementations before release.
Design API specs for LLM-powered agentic clients with discoverable fields, idempotency, dry-runs, semantic descriptions, and safe destructive actions.
Add contract testing from OpenAPI to CI with generated tests, provider checks, consumer expectations, and reliable fixtures.
Specification patterns for event-driven systems: schema versioning, command vs fact events, orchestration choices, idempotent handlers, and replay safety.
Write non-goals that stop scope creep, name deferred work, and give reviewers a clear boundary before implementation starts.
Write payment workflow specs with retryable errors, declined-card handling, timeout behavior, 3DS branches, and dunning states.
Static pagination page · 2/3