THE VERIFIER FOR AI CODING AGENTS

Stop letting AI grade its own homework.

Your agent builds. Test Maze writes the tests first, judges every run with fixed rules and tells the agent exactly what to fix, pinned to the commit. Works with Claude Code, Cursor, Cline and any MCP client.

Free Basic plan Two commands to connect Never reads your repo
~/my-app · claude code + test maze
> Build a password reset flow. Tests first.
SPEAKS MCP · WORKS WITH✦Claude Code✦Cursor✦Cline✦Gemini CLI✦Codex CLI✦Any MCP client
HOW THE MCP WORKS

You ask. Your agent builds. Test Maze keeps score.

Your agent writes the code. The test cases, and the pass or fail, come from Test Maze.

You
Coding agent
Test Maze MCP
Build & verify
“Build the password reset flow.” When it is done, ask the agent to verify it.
Asks for the test cases that verify this feature.
case.list
Returns the test cases, written by you, your QA or your PM.
Runs them, finds the problems and fixes them.
Records the results. Test Maze decides pass or fail, not the agent.
testrun.record_results · pdlc.verify
Pass: ship it. Fail: which case broke, and whether to fix the code or the test.
Write the tests
“Write test cases for checkout: coupons, empty cart, screen readers.” Be as specific as you like.
Asks what is still missing, then saves the new test cases for the next verify.
coverage.gap_for_feature · case.create_batch
Build & verify
  1. You → Coding agent
    “Build the password reset flow.” When it is done, ask the agent to verify it.
  2. Coding agent → Test Maze MCP
    Asks for the test cases that verify this feature.
    case.list
  3. Test Maze MCP → Coding agent
    Returns the test cases, written by you, your QA or your PM.
  4. Coding agent (on its own)
    Runs them, finds the problems and fixes them.
  5. Coding agent → Test Maze MCP
    Records the results. Test Maze decides pass or fail, not the agent.
    testrun.record_results · pdlc.verify
  6. Test Maze MCP → Coding agent
    Pass: ship it. Fail: which case broke, and whether to fix the code or the test.
Write the tests
  1. You → Coding agent
    “Write test cases for checkout: coupons, empty cart, screen readers.” Be as specific as you like.
  2. Coding agent → Test Maze MCP
    Asks what is still missing, then saves the new test cases for the next verify.
    coverage.gap_for_feature · case.create_batch

See it in your workflow

Pick a workflow and watch every tool call travel between you, your agent, Test Maze and your app.

Best for: Vibe coders & founders

“Set up Test Maze for this project”

You ask in plain English.No test framework to learn.

Step 1 / 11
The full loop, step by step
0
lines of your repository read
Your agent sends results, not source.
2
commands to connect
One token per project, from npm.
8
test facets per feature
From happy path to accessibility.
55
MCP tools for your agent
Ask in plain English; it picks.
THE GAP

AI coding agents are great writers. They’re terrible judges.

WITHOUT

The agent grades itself

Self-grading drift
The agent edits the assertion until the test passes. The bug ships.
No shared quality bar
Every agent invents its own idea of “done”.
Hidden flakiness
Retries hide real regressions until the suite is unrunnable.
WITH TEST MAZE

Test Maze is the referee

Tests before code
Acceptance criteria and test cases are on record first, so the target cannot move.
Verdicts by fixed rules
No AI in the pass/fail path. Same results, same verdict.
Pinned to the commit
Every run carries the git sha, branch and a clean-tree flag.
A next step, not a shrug
Fix the code, fix the test, or add coverage, named case by case.
BUILT FOR

Everyone who ships with AI.

Whether you vibe-code a weekend app or run the QA practice for a whole team, Test Maze gives you proof that the AI-built thing actually works.

For vibe coders

You describe it. Your agent builds it. Test Maze proves it works.

  • Tests written from your description, before the code
  • A plain pass or fail, plus what to fix next
  • No test framework to learn. Two commands to connect
See how

For developers

Keep your agent honest: commit-pinned runs and deterministic verdicts.

  • Every result pinned to git sha, branch and clean tree
  • Failure buckets tell the agent code bug vs. test bug
  • Never clones, pulls or scans your repository
See how

For QA & testers

Your test expertise becomes the contract AI-written code must pass.

  • Bring your existing cases in: CSV, PRD or screenshots
  • Explore any running app and turn journeys into cases
  • Suites, plans and runs, manual or Playwright
See how

For product owners

Know what was built, what was tested and what shipped, in plain English.

  • Features, stories and acceptance criteria you can read
  • Releases and success metrics tied to features
  • Every agent action narrated on Agent Sessions
See how

For engineering leads

One verifier and one audit trail across every coding agent on the team.

  • Same quality bar for Claude Code, Cursor, Cline & more
  • Commit-pinned evidence for every verdict
  • Per-workspace tokens, BYOK, self-hostable
See how
YOUR REPO STAYS YOURS

We verify the work. We never read your repo.

Stays on your machine

Your repository.env & secretsYour IDE’s LLMRendered promptsGit diffs & history

Reaches Test Maze

Test casesRun resultsGit sha · branchScreenshots (opt-in)Snippets you ask us to review

Exactly what we see, and what we never do →

WIRE IT UP

Two commands. Your agent now has a verifier.

Create an MCP token in your workspace, then run these in your project folder. The token lands in a git-ignored .env.testmaze.

Full setup guide
terminal · zsh
npx -y @testmaze/mcp init tmt_xxx
claude mcp add tm --scope project \
  -- npx -y @testmaze/mcp

Wire your coding agent
to a real verifier.

Five minutes to install. Deterministic verdicts from the first run. Test Maze never reads your repository — your agent sends results, and verdicts come back. Start on the free Basic plan; upgrade when your team scales.