Developer Guide
Build against a workspace after scope, owners, and credentials are clear.
Developer Guide
Use this section after a workspace owner has approved the feature scope, credential owner, and launch path. It is intentionally compact: use machine-readable contracts for exact endpoint details, and keep human docs focused on implementation choices.
Choose An Entry Point
Authentication & Keys
Choose the right key type, CORS origin, and server boundary.
SDK
Use the TypeScript SDK for browser reads, server writes, and typed workflows.
API
Use OpenAPI and direct HTTP only when the SDK is not the right fit.
MCP
Connect AI agents with approved workspace scope and credentials.
Webhooks
Receive events, verify ownership, and plan retries.
Implementation Order
- Confirm enabled features in Feature Planning.
- Confirm key ownership in Integrations & Keys.
- Start with Authentication & Keys.
- Choose SDK, API, MCP, or Webhooks.
- Return to Launch Readiness before production exposure.
Boundaries
| Need | Use | Avoid |
|---|---|---|
| browser reads | Publishable Key and allowed origins | Secret Key in browser code |
| privileged work | server-side SDK or API calls | user-shared credentials |
| AI-assisted setup | MCP with approved scope | unattended credential changes |
| event processing | webhooks with owner and retry path | business-critical work without observability |
Do not implement against a workspace before the workspace owner approves scope, credential ownership, and launch readiness criteria.
Next Actions
- New integration: open Authentication & Keys.
- Product surface: open SDK.
- Automation or AI handoff: open MCP.