Evaluation kits for karate-agent — the governance layer for AI-built software: requirements, coverage, rules, and evidence as one git-native graph an LLM drives and a human audits. One engine, many surfaces (a JS API + curl + MCP + a served console).
We sent the product sheet, the QUICKSTART, and your
karate.licseparately. Each kit'sREADME.mdsays how to run it — either drop the engine jar (karate-async-<version>.jar) into the kit, or run the one-shotkarate-agentcontainer (no jar to stage). Yourkarate.licgoverns the engine. Learn more: https://karatelabs.io/agent
| kit | what it demonstrates |
|---|---|
policy-api |
one insurance API across three protocols — REST (OpenAPI) + gRPC + Kafka — in a single Coverage report: live probe → durable suite → method coverage → input dimensions → the rich-error path → the gap worklist. |
azure-demo |
git-first requirements traceability for a loan-decision engine — requirements → rules → run evidence in one RTM. The same kit runs ALM-linked on Azure Pipelines (requirement ids click through to Azure DevOps User Stories) and pure-git / spec-driven on GitHub Actions, switched by two env vars. |
The example code in this repository is provided under the MIT License. It is not the
karate-agent license — your karate.lic is sent separately and governs the engine.