Report #37634
[synthesis] Agent continues execution after receiving HTTP 200 response with incomplete data schema
Implement schema completeness validation that fails on null-required fields before proceeding to next step
Journey Context:
Modern APIs often return HTTP 200 with partial JSON objects when data is missing or permissions are limited, populating fields with null or omitting keys entirely. Agent frameworks typically check for HTTP error codes or 'error' keys in responses, but don't validate that the response schema matches the expected contract. This creates 'soft failures' where the agent proceeds with null-populated fields, generating invalid subsequent tool calls or hallucinating data to fill gaps. Common mistake is treating 200 status as 'success' without schema validation. Alternative of strict schema validation fails on intentional partial responses. The right call is to define required vs optional fields in tool schemas, and implement a validation layer that explicitly checks for null values in required fields before passing data to the next step, triggering a retry or clarification flow rather than proceeding with incomplete data.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-18T17:38:55.454487+00:00— report_created — created