I've enhanced Dochia with the ability to generate agent skills. Being a CLI already, it makes it really simple to plug it into agentic workflows.

Run `dochia init-skills` and the coding agent(s) can trigger tests as it builds:

  1. Agent writes endpoint and the OpenAPI spec (or that get's generated from code)
  2. Agent runs: dochia test -c api.yml -s localhost:3000
  3. Dochia produces dochia-summary-report.json + per-endpoint test files
  4. Agent reads errors, fixes code, re-runs
  5. Loop

The JSON output is structured specifically so agents can read and act on it directly, not just humans parsing logs.

It's a native binary (GraalVM), so it's fast on all platforms.

Would love feedback on: is it something you will integrate into your flow, which test playbooks are missing, whether the report format is actually useful in agentic loops, any edge cases you'd expect a tool like this to catch?

GitHub: https://github.com/dochia-dev/dochia-cli Docs: https://docs.dochia.dev

For background, Dochia takes your OpenAPI spec and runs 120+ test playbooks: deterministic negative and boundary scenarios plus chaos testing. No test cases to write, no configuration beyond pointing it at your OpenAPI spec and a running server.