.cursorrules

# WPD Message Gateway — AI agents

**Canonical playbook:** `docs/agents/master-agent.md`

| Intent | Specialist doc |
| ------ | ---------------- |
| Implement, fix, refactor | `docs/agents/delivery-agent.md` |
| Review PR, diff, or design | `docs/agents/review-agent.md` |
| **After every code change** | **`docs/agents/verification.md`** — lint → `/smell develop` → `make audit` |
| **Go work** (`cmd/`, `internal/`, `pkg/`) | **golang-pro** — `.cursor/skills/golang-pro/SKILL.md` |
| **Frontend work** (`frontend/`) | **typescript-react-reviewer** + **typescript-advanced-types** — `.cursor/skills/typescript-react-reviewer/SKILL.md`, `.cursor/skills/typescript-advanced-types/SKILL.md` |
| **Architecture / refactors** | **software-architecture** — `.cursor/skills/software-architecture/SKILL.md` (load with domain skills) |

**Non‑negotiables:** plan before large edits; run **`/smell`** after implementations and before reviews (Clean Code, KISS, DRY, DDD, SOLID + architecture); **`make audit`** before done; fix smell **BLOCKER/HIGH** before PR; never expose secrets; sync docs/Storybook/schema with code; remove dead code and stale comments/docs you touch.

See **`AGENTS.md`** and **`docs/agents/prompts.md`**.