Beyond OpenAPI and AsyncAPI: AI agents wrote the spec, code, and tests - now w…
summary
Agents generate the OpenAPI/AsyncAPI spec, JavaScript logic, and tests, but self‑grading is gamable; a live demo shows a missing input combination (no driver age >70) that leaves requirements red even when endpoints look covered. The setup uses Karate running data‑driven tests, a JavaScript rules engine that converts logic to flowcharts and finds untraversed paths, and agent work by Claude Code plus MCP to close gaps via a minimal coverage scenario add — all run locally to produce a deterministic ship/no‑ship verdict.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:34Karate Labs overview
- 3:39Beyond checkbox testing
- 6:52Rules-based testing
- 9:17Dynamic mocking
- 11:21AI JavaScript for rules
- 13:31Business analyst control
- 15:38Full coverage with less effort
- 17:46Recap of AI workflow
- 19:30Digital twin engine
- 21:13AsyncAPI and gRPC reports
- 23:02Reporting governance demo
- 24:04Local-only AI testing
- 24:46Q&A
related talks
40mAPI Lifecycle: Smarter Design and Governance With Open Tooling
22mI Spoke to My API — Before It Was Even Built
26mMocking and Deploying an AI Chatbot in Minutes
50mSpec-driven Development using Coding Agents
41mGoing beyond copilot with AI Agents
43mDesigning REST APIs for the age of AI agents
30mSpec-Driven Evolution: How AI Agents Are Rewriting the OpenStack/K8s Pla...
24mThe API Economy's Next Chapter: AI Agents as First-Class Clients