CLI
molar verify checks a claim about a page. molar run completes a task. Add --json for scripts and agents.
Let your coding agent test its changes.
Install the molar CLI, sign in once, and check your app in a real browser from your terminal, a CI job or any MCP client. Every check returns JSON your agent can read: pass or fail, the reason, and links to the screenshots and recording.
$ molar verify "the login form is visible" \
--url http://localhost:3000 --json
{
"ok": true,
"pass": false,
"failure_taxonomy": "element_not_found",
"rationale": "The page shows a Sign in button but no password field.",
"hint": "Check that /login renders the form for signed-out visitors.",
"failure": { "code": "element_not_found", "retryable": false },
"execution_mode": "local",
"artifact_manifest_url": "https://api.molar.it/v1/runs/…/artifacts/manifest"
}MCP
Connect Claude Code, Cursor, Codex or any MCP client to https://api.molar.it/mcp.
HTTP API
Start a task with POST /api/v1/agent/tasks and an API token, then poll for the result.
Install and run your first check
Install the CLI, sign in with an API token from the dashboard, then check your local app. Runs against localhost use a browser on your own machine.
- Node.js 18 or later
- Local and hosted runs
- JSON output
Terminal
npm install -g @molar/cli
molar login
molar verify "the login form is visible" \
--url http://localhost:3000 --json$ molar verify "login works" --url http://localhost:3000 --jsontask status · trace referencescreenshots · browser steps inspect and rerunConnect an MCP client
Add Molar as a remote MCP server. It signs in with OAuth, or with an API token. For a local server, run molar mcp serve.
- Tools for tasks, checks, traces and clones
- OAuth 2.1 with PKCE, or an API token
- 100 requests a minute per token
Claude Code
claude mcp add --transport http molar https://api.molar.it/mcp- Completed
- The requested result was observed.
- Failed
- The expected page state was not reached.
- Blocked
- Access, a required input, or a workflow limit prevented progress.
Review the recorded steps and screenshots to see where the task stopped.
Read the result
--json output tells your agent whether the check passed (pass), why (rationale), and where to look (artifact_manifest_url, live_view_url). On a failure it adds a hint and a failure object with a code, a category and whether a retry could help.
Set up a repository
npx molar-agent init adds Molar’s project files to a repository: .molar/config.toml, an AGENTS.md section for coding agents and an MCP config for Cursor. It doesn’t install the CLI. Run npm install -g @molar/cli for that.
Keep your Playwright tests
Molar checks run alongside your existing Playwright suite. Cartographer can also export what it explored as plain Playwright tests that run with npx playwright test, with no Molar account needed.
Frequently asked questions
Have a specific workflow in mind?
Talk to us
Which command should I start with?
Use molar verify to check a claim about a page and molar run for a task. Add --json for agent output.
Where is the hosted MCP endpoint?
https://api.molar.it/mcp. The page at mcp.molar.it has setup steps for each client.
Is there a rate limit?
MCP allows 100 requests a minute per token. During the beta, each organization gets up to 25 hosted runs a day. Runs on localhost use your own machine.
Can I keep Playwright?
Yes. Molar checks run alongside your Playwright tests, and exported tests are plain Playwright files.
Connect your coding agent.
Create a free workspace and describe the first flow to test. Or check a public page first, no account needed.
Free during the beta. No card.