go

Go -- Ship What You Built

Phases 4-6 of /development-lifecycle, standalone command. Use when code written, ready launch.

Assumes: implementation done, tests written. If not -- run /development-lifecycle.

Phase 4: Verify

Run all checks. Fix failures before proceed.

1. bun run type:check (tsgo)
2. bun run lint:fix (biome)
3. bun vitest run --related (changed files)
4. Route touched -> bun vitest run *.browser.test.tsx
5. Dev server running -> browser smoke via scripts/skills-browser.sh (Vercel agent-browser). Skip if not installed.
6. Frontend diff -> run /visual-review (screenshots, states, a11y, console, mobile/cross-browser when feasible). Skip only with reason.
7. When green: commit now. One commit per passing state.

Phase 4b: Refine (Self-Review Loop)

Skip if: trivial change (<10 lines, no logic) | test-only | docs-only.

1. Dispatch self-reviewer agent on session diff
2. Diff >50 lines OR touches auth/security -> also dispatch adversarial-reviewer parallel
3. Process findings by priority -- see REFERENCE.md
4. Fix P0/P1 now, apply P2 safe_auto, show P2 gated_auto to user
5. Commit fixes: refactor(scope): self-review fixes
6. Re-verify (tests + types + lint)
7. Max 2 refine rounds. Then proceed.

Phase 5: Simplify + Ship

1. Run /simplify -- review changed code for reuse, quality, efficiency
2. Fix issues, commit
3. Frontend diff and /visual-review not run this session -> run it now or record explicit skip reason
4. Run /commit-push-pr -- conventional commits, push, open PR
5. Dispatch code-reviewer agent (fresh-eyes review)

Phase 5b: Iterate

1. Monitor: gh pr checks <number> --watch -- stream CI background
2. CI fail -> diagnose, fix, push, re-monitor
3. code-reviewer agent findings -> /resolve-pr-feedback triage, fix, reply, push
4. AI self-review cap: up to 3 auto code-reviewer rounds. Early-exit when reviewer returns APPROVED or empty findings -- never run round N+1 on clean round N. After 3 rounds still noisy -> hand off to human.
5. Human review (incl cloud/Copilot): NO cap. Address EVERY thread. pr-feedback-completeness-stop hook blocks session exit until bash scripts/pr-unresolved-count.sh returns 0 and no CHANGES_REQUESTED reviews remain.

Phase 6: Compound

After non-trivial tasks: "Learn something worth preserve?"

• Write rule to .claude/rules/<topic>.md with paths: glob
• AI bug -> create eval/test fixture catch same error class

Done

1. Post final PR comment: changes, review findings, test coverage
2. Request review: gh pr edit <number> --add-reviewer <username>
3. Report PR URL + CI status
4. Stop. No poll for human approval.

Entry Gate

Before start, check work to ship:

• No uncommitted changes AND no unpushed commits -> nothing do, stop
• On default branch, no feature branch -> auto-spawn via scripts/mux-worktree.sh <type>/<name> before proceed. Never ship from main. [ETHOS: Worktree Isolation]

Skills Composed

Skill	Phase	How
self-reviewer agent	4b	Auto-dispatch on diff
adversarial-reviewer agent	4b	Conditional (>50 lines or auth/security)
/simplify	5	Code quality review
/visual-review	4/5	Frontend screenshot, state, a11y, console, mobile/cross-browser review when UI changed
/commit-push-pr	5	Conventional commits + push + PR
code-reviewer agent	5	Fresh-eyes review on PR
/resolve-pr-feedback	5b	Triage + fix review comments

See REFERENCE.md for detailed checklists, gate logic, flowchart.