Open source · Self-hosted · API release validation

Did the new release break a critical workflow?

MeerGate runs your tests in your own network — from a single API to cross-service end-to-end flows — comparing them with the last known-good results, and delivering the release decision backed by evidence.

Try with Docker Explore the product →

Data never leaves No public endpoint Bring your own AI key

checkout-flow · release gate staging → production

✓

Ready to ship — no new regressions

Order, wallet and payment services verified in one flow.

OrderCreate orderPOST /orders

→

WalletDebit balancePOST /withdrawals

→

PaymentVerify paymentGET /payments/{id}

✓

Endpoint checksauto-generated

38pass

✓

Cross-service flowsyour scenarios

12pass

Known failuresflagged in baseline

2known

50 passed2 known52 checks

0new regressions

50tests passed

3services

82%coverage

The problem

You have test results. Not release confidence.

MeerGate turns scattered test output into the answer to one question: did this change break a critical flow?

Scattered knowledgePostman, scripts and team memory

Critical behavior lives across tools and in people’s heads.

No accessCloud runners can’t reach private services

Order, wallet and payment flows behind a VPN can’t be verified from outside.

False confidence“57 tests passed” isn’t enough

You can’t see what the new release changed versus the last safe behavior.

checkout-flow · release decisionrc.14 → prod

✓

Ready to ship

Order, wallet and payment flows behave the same as the last safe result.

The question MeerGate answers Did this change break a critical workflow?

0New regressions

3Services verified

12Flow evidence

The decision comes not from a single endpoint result, but from cross-service behavior, the baseline diff and execution evidence.

How it works

From API to release decision in four steps

Verify an endpoint, a single service or a critical cross-service workflow with the same model.

01 / CONNECT

Discover your services

Add services via OpenAPI, gRPC reflection or existing endpoint info.

02 / DEFINE

Describe the workflow

Build the expected behavior with natural language, the visual builder or YAML.

03 / RUN

Verify in your own network

Run every step locally, behind a VPN or inside Kubernetes.

04 / DECIDE

See the release risk

Compare the new result with the last safe release: ready, blocked or no baseline.

AI, under human control

AI drafts the flow. You stay in control.

Review, edit and run the generated plan via the visual builder or YAML.

Test Builder · insufficient-balance-order

Run

POSTHTTP · order

Create order

/orders

assert 201extract id

POSTHTTP · wallet

Attempt withdrawal

/withdrawals

assert 400

ASSERTcheck

Verify final state

order == FAILED

balance unchanged

+ HTTP+ gRPC+ WS+ Browser+ Assert

Chat

On insufficient balance the order must be FAILED and the wallet balance must not change.

Verified that correlation_id is propagated between the two services.

One assumption needs review: is the retry behavior idempotent?

Describe a fix or add an assertion… ➤

name: insufficient-balance-order
steps:
  - apiCall: POST /orders
    assert: { status: 201 }
    extract: { start: $.balance }
  - apiCall: POST /withdrawals
    assert: { status: 400 }
  - assert: order == "FAILED"
  - assert: balance == {{ start }}

Why MeerGate?

Measure release risk, not test count

Cross-service flows

Combine order, wallet, payment and browser steps in one scenario.

Compare with the last safe release

Tell known failures apart from new regressions this release introduced.

Run inside a private network

Endpoints, credentials and test data never leave your environment.

Evidence behind the decision

Inspect requests, responses, assertions, diffs and coverage together.

Self-hosted by default

Own your verification infrastructure

Run MeerGate locally, behind a VPN or in your own Kubernetes. Test private services without exposing them to the internet.

Quick setup with Docker Compose

Your own PostgreSQL and secret management

Your own LLM provider or a local model

SSO, RBAC and audit options for enterprise

meergate / quickstart

$ git clone https://github.com/hasimyerli/meergate.git
$ cd meergate
$ make up

✓ api          http://localhost:8080
✓ web          http://localhost:3000
✓ worker       ready
✓ postgres     healthy

MeerGate is ready.

Where MeerGate is going

From verification today to an autonomous release agent

Verification is the foundation. The direction is an agent that runs it for you.

L0 · Manual

You write, run and read API tests yourself.

L1 · AI-assisted

Describe a scenario; AI builds the executable test and you review it.

L2 · Continuous verification

Scheduled reruns, release gates against a baseline, coverage mapping.

You are here

L3 · Release agent

A goal-driven agent plans, runs, interprets and proposes the release decision.

on the roadmap

Ship your next API release with evidence, not guesswork.

From a single endpoint to critical cross-service workflows.

Explore on GitHub ↗