# Measure — MCP server card for AI agents

Get to the root cause of mobile app crashes, errors & slow traces with Measure MCP

Registry name: `sh.measure/measure` · category: Other · version 1.0.0 · updated 2026-09-07

## Connect
- Remote (Claude Code): `claude mcp add --transport http measure https://agent.measure.sh/mcp`

## Facts
- Repository: https://github.com/measure-sh/measure (1,447 stars)
- Website: https://measure.sh/docs/mcp

Before you connect a server: read its repository, and give it only the permissions and keys it needs.

## Links
- AgentGid page: https://agentgid.com/mcp/servers/sh-measure-measure/


## Measure: debug mobile app crashes from your coding assistant

Measure records crashes, errors, slow screens and user sessions from your Android, iOS, Flutter or React Native app. Its MCP server lets a coding assistant such as Claude Code, Codex or Cursor look that data up while it fixes your code.

### Where to use it
- Fix a crash with its stack trace: Ask your assistant for the top crash of the latest version. It reads the error and its stack trace, then opens the right file in your project.
- Which phones a bug hits: Find out whether a crash happens on one OS version, device model or country, so you can test on the right phone.
- Ask about app health in plain words: Ask questions like how many sessions had slow launches yesterday, and get an answer with numbers. Measure's own agent works it out from your data.
- Find where sign-ups drop off: For a sign-up or checkout flow, see which errors, slow requests or network calls users met between two steps before they gave up.
- Slow or failing server calls: List the server addresses your app calls that are slowest or fail most, then compare a version before and after your fix.
- Close bug reports after a fix: Users send bug reports by shaking the phone. Once the fix ships, the assistant marks those reports as closed.

### How to start
- Sign up to Measure Cloud or run it on your own servers, and add the Measure library to your app so data starts coming in.
- Add the server to your coding assistant. Self-hosted copies use their own agent address followed by /mcp.

```
claude mcp add --transport http measure https://agent.measure.sh/mcp
```

- Use any Measure tool; the first call opens a browser window to sign in, later calls work on their own.

### License
Measure is open source under the Apache-2.0 license, so you can run it yourself for free, also in paid products. Measure Cloud has a free plan with the MCP server and a paid Pro plan.

### Good to know
- The tools only find what your app sends. Your app needs the Measure library and a Measure account, hosted or your own.
- On Measure Cloud the plain-question tool belongs to the Pro plan and is billed at the model provider's token price plus a markup.
- Changes to funnels apply to the whole team and need a role that may edit funnels. Bug report status can also be changed.
- On your own servers the question-answering agent is off until you turn it on and add a model provider key; the other tools work without it.

Checked 2026-11-10 against: https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/README.md, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/LICENSE, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/frontend/dashboard/content/docs/mcp.mdx, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/frontend/dashboard/content/docs/hosting/agent.mdx, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/backend/agent/mcp/mcp.go, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/backend/agent/mcp/analytics.go, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/backend/agent/agent/toolbox.go, https://github.com/measure-sh/measure/blob/3141abe428514e943d980cc8350097c9a7c2b526/backend/agent/main.go, https://github.com/measure-sh/measure/releases.atom, https://measure.sh/docs/mcp, https://measure.sh/docs/agent, https://measure.sh/pricing, https://measure.sh/privacy-policy, https://measure.sh/llms.txt

### Mistakes to avoid
- Don't: writing a filter_expr from guessed key names → Do: call get_filter_keys for that kind of data first, then get_filter_values for a key's values (The keys and their operators differ per kind of data.)
- Don't: a follow-up ask_question such as "and on iOS?" → Do: repeat the app ids, the earlier findings and numbers, and the answer's Method line in the new question (Each ask_question call starts fresh and remembers nothing from the last one.)
- Don't: asking ask_question to create a funnel or close a bug report → Do: create_funnel, update_funnel or update_bug_report_status (The agent behind ask_question only reads data.)
- Don't: "url" or "httpUrl" in the Antigravity config → Do: "serverUrl": "https://agent.measure.sh/mcp" (Antigravity reads only serverUrl for remote servers.)

### Questions
- Does my assistant need a key? No. On the first tool call your client opens a browser window to sign in to Measure; later calls work on their own.
- Is the MCP server on the free plan? Yes. The free plan includes the MCP server; the Pro plan adds Measure Agent, which answers plain questions.
- Which assistants does it work with? The docs show setups for Claude Code, Codex, Cursor, VS Code, Claude Desktop, Android Studio, Antigravity, Zed and opencode; any client with remote servers over streamable HTTP can connect.
- Which app platforms does Measure cover? Android, iOS and iPadOS, Kotlin Multiplatform, Flutter and React Native.

---
AgentGid's catalog is curated: every entry was checked against the maker's official pages (registry, repository, license and pricing pages) on the date shown; a license value shows the quote it stands on wherever AgentGid has one. Usage numbers come from AgentGid's own scan of public repositories set up for AI coding agents. Claims by makers are reported, not tested by AgentGid.
