plan-eng-review · git:20260926.941aa31 · 2026-09-26 · sha256 03eab112ed564185
plan-eng-review git:20260926.941aa31A
Immutable. This exact content never changes and is served at /api/v1/blob/03eab112ed564185.
--- name: plan-eng-review description: Evaluates system architecture, failure modes, data modeling, concurrency, rate limits, testing strategy, error handling, and API contract boundaries from a Principal/Staff Engineer perspective. --- # Plan Engineering Review ## Overview A robust product requires rock-solid engineering foundations. `plan-eng-review` stress-tests technical designs before code is written, uncovering hidden edge cases, race conditions, N+1 queries, unhandled external API failures, and missing error boundaries. ## When to Use - After CEO and Design reviews, before drafting atomic tasks or implementation. - When designing database schemas, backend services, or complex API endpoints. - When integrating with external AI APIs, payment webhooks, or third-party OAuth. ## Principal Engineer Review Checklist ### 1. Data Integrity & Schema Design - Proper normalization vs. deliberate denormalization. - Foreign keys, cascading behaviors, indexing on query filters/sorts. - Idempotency keys on mutations and financial transactions. ### 2. Failure Modes & Resilience - What happens when OpenAI / Anthropic / Stripe returns 503 or 429? - Backoff, retry policies, exponential jitter, circuit breakers, fallback models. - Graceful degradation in the UI when background jobs are queued or delayed. ### 3. Concurrency & Rate Limiting - Race condition mitigation (e.g., atomic DB updates, Redis locks). - API token bucket / leaky bucket per IP/User. ### 4. Testing & Verification Pyramid - Unit tests for pure business logic & parsers. - Integration tests for database queries and webhook processors. - End-to-end sanity tests for critical user conversion paths. ## Output Format Save the Engineering review in `brain/<conversation-id>/eng_review.md`: ```markdown # Engineering Review: [Architecture/Feature Name] ## 1. Architectural Architecture Summary - **Data Flow**: Client -> [API/Edge] -> [Service Layer] -> [DB / Vector Store / Cache] - **Key Abstractions**: [Services, Repositories, MCP Tools] ## 2. Risk Matrix & Mitigations | Failure Mode | Likelihood | Impact | Mitigation Strategy | | :--- | :---: | :---: | :--- | | External API 429/500 | Medium | High | Exponential backoff + fallback model | | Double Webhook Delivery | High | Critical | Idempotency table check | | Slow Vector Search Query | Medium | Medium | HNSW index + Redis query cache | ## 3. Testing & Verification Gates - [Unit test suite paths] - [Integration test plan] ## 4. Engineering Verdict - [PASSED TO SPEC-DRIVEN IMPLEMENTATION / REQUIRES SCHEMA REVISION] ```