01 / DECIDE
Architecture decision
Use it to connect governed external capabilities or data without rebuilding every integration as a custom function.
Inventory server identity, transport, authentication, exposed tools, data returned, approvals, and failure behavior before enabling it. Map input, output, state, and side effects as one observable path before deciding which layer owns treat mcp servers as external principals.
02 / BUILD
Three-step implementation
- 01
Frame the contract
Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it to connect governed external capabilities or data without rebuilding every integration as a custom function.
- 02
Build one narrow path
Implement one end-to-end path with request correlation, typed state, and reversible failure handling. Inventory server identity, transport, authentication, exposed tools, data returned, approvals, and failure behavior before enabling it.
- 03
Prove the outcome
Turn acceptance into a repeatable fixture, contract test, or browser test. Negative tests prove an untrusted server, changed tool schema, expired credential, and injected content fail safely.
03 / BOUND
Production boundary
Only connect trusted servers, minimize tools, validate returned content, and keep high-impact calls behind approval.
04 / PROVE
Acceptance evidence
Negative tests prove an untrusted server, changed tool schema, expired credential, and injected content fail safely.
SOURCE / HTTP
Reproducible source probe
curl -fsSI 'https://developers.openai.com/api/docs/guides/tools-connectors-mcp' | sed -n '1,5p'