01 / DECIDE
Architecture decision
Use it before scaling agent workflows or changing models, prompts, tools, and orchestration in production.
Connect run, turn, model, tool, handoff, approval, latency, usage, and terminal outcome with stable correlation identifiers. Map input, output, state, and side effects as one observable path before deciding which layer owns trace decisions without logging secrets.
02 / BUILD
Three-step implementation
- 01
Frame the contract
Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it before scaling agent workflows or changing models, prompts, tools, and orchestration in production.
- 02
Build one narrow path
Implement one end-to-end path with request correlation, typed state, and reversible failure handling. Connect run, turn, model, tool, handoff, approval, latency, usage, and terminal outcome with stable correlation identifiers.
- 03
Prove the outcome
Turn acceptance into a repeatable fixture, contract test, or browser test. An incident can be reconstructed from identifiers and outcomes without needing unrestricted raw prompt access.
03 / BOUND
Production boundary
Redact credentials and sensitive content, define retention, and separate operational metrics from full conversation storage.
04 / PROVE
Acceptance evidence
An incident can be reconstructed from identifiers and outcomes without needing unrestricted raw prompt access.
SOURCE / HTTP
Reproducible source probe
curl -fsSI 'https://developers.openai.com/api/docs/guides/agents/integrations-observability' | sed -n '1,5p'