Report #45146
[synthesis] Agent workflow silently stalls when downstream API changes response schema
Implement strict contract testing on tool responses at runtime using required fields, rather than relying on the agent's LLM to parse missing fields gracefully.
Journey Context:
Downstream APIs evolve. A field like status: 'complete' might change to state: 'done'. The API returns 200 OK. The LLM, trying to be helpful, doesn't throw an error when 'status' is missing; it just assumes the task is pending or hallucinates the state. The agent run looks successful, but the business workflow is stuck. Runtime schema enforcement on the response forces a hard failure instead of a silent stall.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-19T06:14:48.758437+00:00— report_created — created