Built for LLM agents.
Works great for humans.
Declarative end-to-end testing across HTTP, SQL, Kafka, and MongoDB. Write tests as pure data. Run them from the CLI or let your AI agent handle it.
{:description "Register and verify" :steps [{:set {:email :gen/email}} {:call :api/register :body {:email :ctx/email} :expect [[:status 201]] :save {:token [:body :token]}} {:call :api/profile :headers {"Authorization" :ctx/token} :expect [[:status 200] [:body :email :ctx/email]]}]}
HTTP · SQL · Kafka · MongoDB · Util
Zero exploration
Every response guides the next step. No docs to read, no APIs to discover. AI agents reach a passing test in 4 commands. Humans love it too.
$ zriz resources→ zriz resource myshop-apiSelf-correcting tests
Wrong field name? You get the right one suggested and a one-command fix. No more guessing what the API actually returns.
did-you-mean: user_id→ zriz validate --fixVersion-controlled infrastructure
Test resources live in your repo as data files. See what changed before you deploy. Apply with one command.
$ zriz diff api.edn$ zriz apply api.edn5
Protocols
19
CLI Commands
21
Generators
22
Assertions
11
MCP Tools