quality assurance for coding agents

Dino compresses hours of API testing, security checks, and production hardening into minutes.

Free tier forever.

Backed by

why dino

Coding is getting faster. Trust isn't.

AI agents change software in seconds. But the agent that wrote the code shouldn't be the only thing deciding it's safe to ship. Dino sits outside the generation loop: the agent builds, Dino verifies the running API, and returns a verdict it can act on.

Without Dino
> agent: shipped the API change
  ✓ looks correct
  ✓ tests pass

# three days later, in production
  ✗ /orders returns 200 with a null total
  ✗ a deprecated field is still served
  ✗ an auth boundary was silently dropped
With Dino
> dino scan --endpoint https://api… --fail-on-high
  verdict: At risk      exit code 3
  ✗ HIGH  orders.total     nullable, schema says non-null
  ✗ HIGH  auth boundary    role bypass on 2 operations
  ✗ MED   billing.plan     deprecated field still served

> agent: fixed. re-running.
  verdict: Healthy      exit code 0   → safe to ship

what you get

Move fast. Ship what actually works.

  • Ship with confidence

    Merge the changes your agents write, knowing the running API still works.

  • Catch breaks before customers

    Dino flags the change that breaks your consumers before it reaches production.

  • No QA to maintain

    Point Dino at your API. There's no test suite to write or keep green.

faq

Questions

Does Dino replace my coding agent?

No. Dino sits alongside whatever agent you already use, Claude Code, Codex, Cursor, or a local model. The agent builds the change; Dino tests the running API and hands back a verdict the agent can act on. It never writes your code, it verifies it.

What does “deterministic” actually mean here?

Same evidence and same policy always produce the same verdict. Dino turns observed behaviour into evidence and evaluates it against fixed rules, so results are reproducible in CI, not a best-effort opinion that varies between runs.

Does it work with REST and GraphQL?

Yes. Point Dino at a running endpoint or a spec URL and it works with both. Run dino init then dino scan to get a verdict, with no test suite to write or maintain.

How does Dino know what broke?

Dino remembers what your API looked like before the change: operations, schemas, findings, and behaviour. Every future run compares against that known-good state, so it can tell a long-standing behaviour from something your latest change just broke.

Is it free?

Yes. There's a free tier forever, so you can point the CLI at your own API and get a verdict at no cost.

// SHIP

What's New

v1.0.7

14 releases shipped

Read full changelog

Give every coding agent
a quality assurance layer.

Dino is the deterministic verification layer for APIs. Agents build the software. Dino proves it works.

Works with Claude Code, Codex, Cursor, and CI.