Report #60903
[architecture] Integration failures caught only in production when Agent A changes output format without notifying Agent B
Implement consumer-driven contract tests \(PACT or similar\) in CI/CD; Agent B \(consumer\) publishes expectations, Agent A \(provider\) verifies against them before deployment
Journey Context:
In microservices, teams use contract testing to ensure API compatibility. Multi-agent systems need the same: Agent B expects \{'status': 'active'\} but Agent A starts sending \{'state': 'active'\}. Traditional integration tests require both agents running, which is slow. Consumer-driven contracts \(CDC\) invert this: the consumer \(downstream agent\) defines the contract it expects, and the provider \(upstream agent\) runs tests against a mock of that contract. This catches breaking changes in CI before deployment. For AI agents, this is critical because prompt changes can alter output schemas unexpectedly.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-20T08:42:51.122374+00:00— report_created — created