preparing-releaseslisted
Install: claude install-skill JavierQuinan/ai-dev-team-marketplace
# Preparing releases
Produce a release-readiness verdict backed by checks actually run in this session — never assert readiness from inspection alone. See [enforcing-safety-baseline](../enforcing-safety-baseline/SKILL.md): deploying to production always requires explicit human confirmation, this skill prepares and verifies, it does not deploy.
## Checklist
Run each applicable check and record its real result:
1. **Git state** — clean working tree or explicitly-reviewed pending changes; correct target branch; no accidental local-only commits.
2. **Build** — the project's actual build command, executed, output inspected.
3. **Typecheck** — if the stack has one (TS, mypy, etc.), executed.
4. **Lint** — executed, not assumed clean.
5. **Tests** — unit/integration suite executed; report pass/fail counts, not "should pass."
6. **E2E** — relevant Playwright/Cypress suite executed if one exists and the change touches user-facing flows.
7. **Secrets** — no `.env` with real values, no hardcoded credentials/API keys/tokens in the diff or tracked files; `.env.example` uses placeholders only.
8. **Demo/tenant data contamination** — no test/seed/demo data, debug flags, or hardcoded tenant-specific values left enabled for production.
9. **Migrations** — reviewed for safety (reversibility, lock behavior on large tables, backwards compatibility during rollout); not auto-applied to production by this skill.
10. **Env examples** — `.env.example` (or equivalent) reflects the actual required