FinanceGPT Developer Platform
Integration certification standard
A transparent evidence standard for human review of FinanceGPT integrations. Eligibility is not certification, production approval or financial authority.
Certification evidence requirements
| Requirement | Required | Evidence |
|---|---|---|
| Pinned API contract | Yes | Current OpenAPI SHA-256 and compatibility policy are explicit and aligned. |
| Least-privilege scope alignment | Yes | Developer Catalog and OpenAPI scope requirements have no current mismatch. |
| Sandbox validation | Yes | An active sandbox application and recent governed sandbox execution evidence exist. |
| Operational readiness | Yes | Current readiness controls have no blocking failure. |
| Webhook delivery posture | when used | Active webhook integrations have no unresolved failed/retrying delivery posture in the current review window. |
| Support diagnostics | Yes | A redacted workspace-scoped diagnostic evidence packet can be produced without secret or payload disclosure. |
| Authority separation | Yes | API identity, model choice and certification evidence do not create model-promotion, QLM-rebinding or Financial Actions authority. |
Human issuance boundary
FinanceGPT can calculate whether a workspace is eligible for human certification review. Certification readiness does not issue a certificate automatically, activate production, broaden scopes, create credentials, promote models, rebind QLMs, change billing, or approve or execute a Financial Action.