Connector: Design Integration

Prompt file imported from Cataldir/governed-second-brain (.github/prompts/connector-design.prompt.md). Copyright stays with the author.

Coordinate enterprise connector design:

1. API Discovery — Invoke enterprise-connectors via #runSubagent to:

   • Analyze the target platform's API surface (endpoints, auth model, rate limits, pagination)
   • Identify the minimum API scope required for the use case
   • Document data models and field mappings (source → internal → target)
   • Classify the platform category (PIM, DAM, CRM, Commerce, Inventory/SCM, Data/Analytics, Integration, Workforce, Identity, Privacy)

2. Architecture Design — Invoke SystemArchitect via #runSubagent to:

   • Define the connector's place in the overall integration architecture
   • Choose communication pattern (sync request-response, async events, batch ETL, webhook-driven)
   • Design error handling strategy (retry policies, dead-letter queues, circuit breakers)
   • Specify contract versioning approach for the connector API surface

3. Implementation Plan — Invoke the appropriate language specialist via #runSubagent:

   • PythonDeveloper — For Python async connectors (httpx, aiohttp)
   • RustDeveloper — For high-throughput connectors (reqwest, tokio)
   • TypeScriptDeveloper — For Node.js connectors (fetch, graphql-request)
   • Define: auth flow, pagination handling, rate limit respect, data transformation pipeline

4. Security Review — Invoke PlatformEngineer via #runSubagent to verify:

   • Credential storage (Key Vault / env vars, never hardcoded)
   • OAuth token refresh handling
   • Input validation on all external data
   • Audit logging for compliance

5. Testing Strategy — Define:

   • Contract tests against the external API (recorded fixtures / WireMock)
   • Integration tests with sandbox environment
   • Load tests to validate rate limit handling

Deliver a connector design document with architecture diagram, data mapping table, implementation spec, and test plan.