v0.8.0 is out. See what is new

Mock APIs. Simulate dependencies. Build without blockers.

Mock the APIs you do not have, test the ones you do. One small program on your machine, with no account and no cloud.

curl -fsSL https://www.zyvra.dev/install.sh | sh

Installs for your user only, no sudo. How the download is verified

FreemacOS, Linux, WindowsPlain files, Git-friendlySigned releases
zyvra.yamlplain file, lives in Git
- method: GET
  path: /products/{id}
  rules:
    - name: unknown product
      when:
        - { source: path, name: id, op: equals, value: "999" }
      respond: missing
  scenarios:
    found:   { status: 200, body: { name: Keyboard, price: 49 } }
    missing: { status: 404, body: { error: product not found } }
terminal
$ curl -i localhost:18081/products/999
HTTP/1.1 404 Not Found
X-Zyvra-Rule: unknown product
X-Zyvra-Scenario: missing

{"error": "product not found"}
See it work

One minute, from request to green tests

A mock shop API, the API client, the Console, Traffic and a collection run.

A request to a mock shop API, a product that does not exist (a rule answers 404), the Console and Traffic views, and a whole collection run with its checks.
What it does

From "the API isn't ready" to "the tests are green"

Use the mock side, the client side, or both in one project.

Mock

A mock that behaves like the real thing

Stand in for a service that is not ready, is slow, or that you would rather not call. Rules, scenarios and state decide the answer from the request itself.

  • Rules pick a response by path, query, header or body; the first match wins
  • Scenarios flip an endpoint between happy path, errors and edge cases
  • Webhooks, stored state and scripted answers
  • Start from an OpenAPI spec, a recording or a collection, and keep it in step when the spec changes
The endpoint editor for POST /orders showing two rules: a test card number gets a decline, an amount above 1000 gets an in-review answer
Break

Make it fail on purpose, the same way every time

Find out what your client does when the other side is slow, down, rate limited or wrong, without waiting for it to happen.

  • Latency, errors, dropped connections, cut-off bodies and rate limits with Retry-After
  • Fail a share of requests, every 3rd one, or only when a header asks for it
  • Ready-made situations: Slow 3G, a flaky mobile connection, goes down after five requests
The Faults tab of POST /orders: a slow response and a 503 error, each applying only when an x-chaos header is sent
Test

Send a request, see exactly what answered

A full API client that knows your mocks: every response says which service, rule and scenario produced it, and whether it matches your spec.

  • REST, SOAP and GraphQL, with collections, folders and shared auth
  • Environments, variables, cookies, cURL import and code snippets
  • Basic, Bearer, API key, JWT, OAuth, client certificates
  • Import from Postman, OpenAPI, cURL or a browser recording; export back to Postman
The API Client after GET /products/999: a 404, a note that it matches the shop spec, and that mock shop-api answered through the rule "unknown product"
Inspect

See what was really sent

The Console logs every request the dashboard sends, from the API Client, the Test Runner and workflows alike, with real values in place of your {{variables}}.

  • Headers, body, auth and timings for what went out and what came back
  • Secrets stay masked until you ask to see them
  • A request that never got an answer is logged too, with the reason
The Console open under the API Client, listing requests from the Test Runner and the API Client with their status codes and timings
Automate

Turn requests into tests, run them anywhere

Add checks to a request, then run the whole collection here, or in CI with one command and a JUnit report.

  • Checks on status, headers, JSON paths and JSON Schema
  • JavaScript test scripts, with a Postman-compatible pm.* layer
  • Data-driven runs from a CSV or JSON file, and exit codes your pipeline understands
The Test Runner after running the Shop API collection: ten requests passed, thirteen checks
Observe

Never wonder why a mock answered that way

Every request a mock receives is recorded with the rule and scenario that handled it, so a surprising answer is easy to explain.

  • A live stream with headers and bodies, narrowed to errors or to requests no route matched
  • "Why this response": which rule matched, which did not, and the values each one saw
  • Create a rule straight from a captured request
The Traffic list: requests to the shop services with their status codes, a count of failed requests, and filters for service, method and outcome
And OpenAPISOAP and WSDLGraphQLPostman importcURL and HAR importWorkflowsScenariosOAuth and JWTClient certificatesJUnit reportsCSV-driven runs
Coming from somewhere else?

Pick the tool that fits. Here is where Zyvra does.

Every tool here is good at something. These are the honest trade-offs.

If you use Postman

Keep your collections. Drop the account.

Import collections, environments and globals, and run mocks and tests on your own machine. Nothing goes to a cloud.

Be aware: No team workspaces: Git is how you share.

Zyvra as a Postman alternative
If you use WireMock

Mocks as files, with a dashboard.

Rules, scenarios and failures from a UI or plain YAML, plus an API client and a test runner in the same program.

Be aware: Younger, and not a JVM library.

Mock API tools compared
If you use Mockoon, json-server

More than routes and JSON.

Stored state, webhooks, failure modes and tests in CI, when a fixed response is no longer enough.

Be aware: More to learn than a one-file fake.

When each tool fits
If you use Hosted mocks

Local by default.

Your mocks and your data stay on your computer, work offline and cost nothing.

Be aware: No public URL to hand to outsiders.

How Zyvra mocks an API
Actively built

Actively built, steadily shipped

Latest release: v0.8.0, November 18, 2026. 24 releases so far, each one signed.

v0.8.0
  • GraphQL
  • A GraphQL request
  • A Schema panel
  • Mistakes are underlined
v0.7.0
  • SOAP and WSDL
  • Import a WSDL
  • REST or SOAP, never both
  • Writing SOAP mocks
v0.6.0
  • Workflows: an Evaluate block runs JavaScript

Read the full changelog

Built to be trusted

You should not have to take our word for it

Most of this can be checked.

It runs on your machine

Everything listens on 127.0.0.1. Nothing is reachable from other computers, and there is no cloud service behind it.

Your projects are plain files

YAML in your own folder. Read them, review them, keep them in Git. Nothing is locked in a database you cannot open.

Secrets are encrypted, and out of the files

Mark a variable secret and its value is stored encrypted outside the project, then masked in responses, history and traffic.

Scripts only run if you allow them

JavaScript is code, so it stays off until you trust the project. It runs in a sandbox with no file, network or process access.

Releases are signed

Every release is signed with Zyvra's own key. Zyvra refuses an update the key did not sign, and never installs one without asking.

No account, no tracking

Zyvra reports nothing about you or your projects. The optional daily update check sends only a version, a system and an architecture, which this website only counts: no address, no identifier.

Try it in two minutes

Install it, create a project, and send your first request to a mock. There is nothing to sign up for.