
Endform MCP Server: Setup for Claude Code, Cursor and Other Apps
Lets Claude, Cursor and Codex find failing, flaky and slow Playwright tests on Endform, and debug them live.
Endform runs your Playwright tests in parallel on its own servers. Its MCP server lets your assistant look through your test history, set quarantine rules and pause a test mid-run to inspect the page.
The server is part of Endform's own command line tool and runs on your computer. Sign in with the Endform login command first, or set an API key. Start it from your project folder so it can find your Playwright config.
Good for
- Teams that run their Playwright suite on Endform and want faster triage
- Developers chasing a test that only fails in CI
- QA leads who quarantine flaky tests so they stop blocking pull requests
Works in
12 should work
Tap an app to see how to set it up and any known problems.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Supports local MCP servers.
Client does not support this transport.
Client does not support this transport.
Install Endform MCP
Claude Code
- 1
Copy this prompt.
Add the Endform MCP server to Claude Code. Run this command: claude mcp add --transport stdio --scope local endform -- npx -y endform@latest mcp Then check that it loaded and tell me if I need to sign in or restart.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
Requires Node.js (npx), An Endform account: run npx endform@latest login first, or set ENDFORM_API_KEY.
Run npx endform@latest login once first, and start the server from your project folder so it finds your Playwright config.
From the vendor's docs. Source.
Tools
Charts pass rates, run counts and durations
Compares test or suite runs over time and returns a time series. You choose the metric, a grouping such as branch or outcome, a filter and a period, which defaults to 7 days.
Try: “What was our pass rate on main over the last 30 days?”
Network ·
query_analyticsLists real test runs that match a filter
Returns recent suite or test runs with project, outcome, duration and every attempt, including error messages and trace IDs. It returns up to 50 runs per call.
Try: “Show the last 10 failed runs of the checkout test.”
Network ·
query_samplesReads the detailed trace of one test attempt
Returns the raw trace for a test attempt: Playwright actions, network activity and timings. You need the trace ID from a sample.
Try: “Why was this attempt slow? Look at its trace.”
Network ·
get_otel_traceLists active quarantine rules
Returns active rules with their IDs, scope, reason and expiry. It can include expired and deactivated rules.
Try: “Which tests are quarantined right now?”
Network ·
list_quarantine_rulesQuarantines a flaky test (or a group)
Creates a rule for a suite, optionally narrowed to one test, project or branch, with a reason and an expiry. Quarantined tests stop blocking your runs.
Try: “Quarantine the flaky search test on main until Friday.”
Write ·
create_quarantine_ruleChanges an active quarantine rule
Replaces the editable settings of an active rule while keeping its ID and priority. You must supply every field.
Try: “Extend the search quarantine by a week.”
Write ·
update_quarantine_ruleTurns a quarantine rule off
Deactivates one rule by its ID so the matching tests count again.
Try: “Remove the quarantine on the search test.”
Write ·
deactivate_quarantine_ruleStarts a paused debugging session for one test
Starts a remote session that pauses inside exactly one Playwright test and returns a session ID and a snapshot of the page. Startup can take a few minutes, and sessions end after 15 minutes.
Try: “Start a live session for tests/checkout.spec.ts:47.”
Runs code ·
live_test_startRuns code inside the paused test
Runs JavaScript in the paused test's scope, with the page and fixtures available. The assistant can inspect locators, read state, click through the page or take screenshots.
Try: “Click the pay button and tell me what the page shows.”
Runs code ·
live_test_runChecks the state of a live session
Reports whether the session is starting, paused, running, ended or stale, with a fresh page snapshot when paused.
Try: “Is the live session still paused?”
Network ·
live_test_statusEnds a live session and cleans up
Stops the session and removes its local and remote resources. The assistant should call it when debugging is done.
Try: “Stop the live session.”
Runs code ·
live_test_stopUsed by the assistant behind the scenes
Lists your Endform organizations
Shows the organizations your Endform login can use, so the assistant can pick one when a tool asks.
Try: “Which Endform organizations do I belong to?”
Network ·
list_organizations
From the official docs
Permissions
medium risk · 3 things to watch
Medium risk
The assistant can run code against a live test on Endform and change which tests are quarantined. Use it on test environments, and give quarantine rules an expiry.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
It runs code in a live test on Endform
The live test tool runs JavaScript in a paused Playwright test on Endform's servers, so it can click, type and read the page. Point it at test environments, not production accounts. endform.dev
It sends your tests to Endform
Endform says the CLI sends the Playwright tests you select, what they need to run and their results to Endform. Read its privacy policy before using private code. github.com
Quarantine rules hide failing tests
A rule makes matching tests stop blocking your runs. Give rules an end date and review them. endform.dev
Security advisories
No published security advisories as of Nov 1, 2026.
Technical details
| Turning on read-only | No read-only mode is documented. |
|---|---|
| Talks to | Endform (analytics, quarantine rules and remote live sessions) |
| Sign-in and keys | Sign in once with the Endform login command, or set ENDFORM_API_KEY in the server's environment. The docs say not to commit keys in a project-level config. |
| Scopes |
|
Questions
Is the Endform MCP server official?
Yes. Endform ships it inside its own command line tool, and the repository belongs to the endformdev organisation.
Do I need an Endform account?
Yes. Run the Endform login command first, or set an API key in the server's environment for non-interactive use.
Where do I run it from?
From your project folder. The debugging tools need to find your Playwright config and tests.
Is this the same as the Endform skill or plugin?
They are related. The Claude Code, Codex and VS Code plugins add this server for you, and the skill teaches the agent when to use it. You can also add the server alone.
Can it change or delete anything?
It can create, change and deactivate quarantine rules, and run code in a live test session. It has no tool that deletes tests or runs.
How long does a live debug session last?
Up to 15 minutes. Starting one can take a few minutes, and the assistant should stop it when done.
Does it work without Playwright?
No. It is built for Playwright test suites that run on Endform.
Information
updated Nov 2, 2026
- Publisher
- Official
- Package
endform- Registry
- npm
- Transport
- stdio
- Auth
- API key
- Version
- 0.82.0
- License
- MIT
- Last commit
- Oct 28, 2026
- Open issues
- 0
- Listed in
- Official MCP registry
Sources
Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.
Show all 2Hide
- endform.dev · endform.dev
- github.com · github.com