01 / DECIDE
Architecture decision
Use it when an application needs answers grounded in uploaded manuals, policies, cases, or project artifacts.
A hosted retrieval tool still requires corpus governance: ingestion status, metadata, access filters, citations, refresh, and deletion. Map input, output, state, and side effects as one observable path before deciding which layer owns build file search with ownership and freshness.
02 / BUILD
Three-step implementation
- 01
Frame the contract
Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it when an application needs answers grounded in uploaded manuals, policies, cases, or project artifacts.
- 02
Build one narrow path
Implement one end-to-end path with request correlation, typed state, and reversible failure handling. A hosted retrieval tool still requires corpus governance: ingestion status, metadata, access filters, citations, refresh, and deletion.
- 03
Prove the outcome
Turn acceptance into a repeatable fixture, contract test, or browser test. A tenant-isolation suite proves cross-user queries cannot retrieve titles, snippets, identifiers, or content.
03 / BOUND
Production boundary
Authorization must be evaluated before retrieval, and deleted documents must stop influencing answers and citations.
04 / PROVE
Acceptance evidence
A tenant-isolation suite proves cross-user queries cannot retrieve titles, snippets, identifiers, or content.
SOURCE / HTTP
Reproducible source probe
curl -fsSI 'https://developers.openai.com/api/docs/guides/tools' | sed -n '1,5p'