Ask Claude about your AI search visibility
Connect Claude Code, the Claude desktop or web app, or another MCP client with one URL and a sign-in. Claude then answers from your tracked prompts, citations, competitors, audits and connected analytics.
Add Measure to Claude Code or the Claude desktop and web app, sign in once, then ask Claude about your AI-search performance.
Settings → Connectors → Add custom connector → paste the Server URL
- Claude Codesince 9/22/2026 · 1 sessionConnected via Measure
- Claudesince 9/29/2026 · 2 sessionsConnected via Measure
In industry data (ChatGPT and AI Overviews), citations are up 6 on last week to 46, and Acme is newly cited on 2 questions. One matches your tracked prompt “expense tracking software for agencies”, where your own scans name Acme in 27 of 48 measured answers over the last 30 days (56%). One answer from each of 4 engines costs 10 credits to re-check it live (1,840 left); run it from the Measure app or Agent M.
Product preview, illustrative data
What it does
One URL and a browser sign-in
Add https://www.measure.co/api/mcp to Claude as a custom connector, or run one claude mcp add command. You sign in with Measure over OAuth 2.1, so there is no API key to copy, store or rotate.
44 tools, 43 shared with Agent M
Visibility, tracked prompts, citations, competitors, content gaps, site audits, technical checks and your connected GA4 and Search Console data. The MCP server and Agent M share one tool catalog, so both read the same numbers.
Scoped by your sign-in
The OAuth token decides which workspace Claude can see; tool arguments never can. Claude works on your primary project by default; list_projects shows the other projects in that workspace, and any tool can run on one of them.
Cost estimates before any scan
Live scans and new tracked prompts stay in the app. Claude can call estimate_scan_cost to price a scan engine by engine against your remaining credits, and you run it from Measure when you are ready.
Shared memory and saved reports
Claude reads and updates the same project memory as the dashboard and Agent M, and can save a finished HTML report to your Reports page for the team to open, share or print.
Workflows Claude can follow
list_skills loads Measure's step-by-step workflows for an AI visibility audit, closing citation gaps, reviewing your prompt portfolio and writing a report, so each run starts from the same checklist.
Three steps, start to finish
Step 1: Add the server URL
In the Claude desktop or web app, open Settings, then Connectors, then Add custom connector, and paste the URL. In Claude Code, run the command shown under Settings & MCP.
Step 2: Sign in with Measure
Your client opens a Measure sign-in page in the browser. Authorize it for your workspace. Connected clients are listed in Settings & MCP, where you can disconnect any of them.
Step 3: Ask in plain language
Ask who is ahead of you, which prompts are weakest or where your content gaps are. Claude picks the tools, and the server tells it to say whether a number comes from your scans or industry data.
The specifics
Engines, data sources, cadence and plan limits, stated as the product works today.
- Endpoint
- https://www.measure.co/api/mcp over Streamable HTTP and JSON-RPC 2.0. MCP protocol 2025-06-18, with 2025-03-26 and 2024-11-05 also accepted
- Sign-in
- OAuth 2.1 with dynamic client registration and PKCE. No static API keys and no REST API. Disconnect a client any time in Settings & MCP
- Tools
- 44: 43 shared with Agent M plus list_projects. Live scans and adding prompts run only in the app and Agent M; web search is Agent M only
- Rate limits
- Tool calls per workspace per minute: Starter 60, Pro 120, Advanced 300, Enterprise 1,200. Audit tools have their own lower limit
- Credits
- Most tools use none. Industry-data and Google top 10 lookups are free from cache; a cache miss costs the same 1 to 45 credits as in the app
- Plans
- Included on Starter, Pro, Advanced and Enterprise
Common questions
Which AI clients can connect to the Measure MCP server?
Claude Code connects with one command, claude mcp add --transport http measure https://www.measure.co/api/mcp, then /mcp to sign in. In the Claude desktop and web app, add the same URL under Settings, Connectors, Add custom connector. Other MCP clients that support remote servers over Streamable HTTP with OAuth sign-in can use the same URL.
Do I need an API key to use Measure in Claude?
No. The Measure MCP server uses OAuth 2.1 with dynamic client registration and PKCE: your client opens a Measure sign-in page and you authorize it for your workspace. Measure has no static API keys and no REST API, so the MCP server is the only programmatic way to reach your data. You can disconnect any client in Settings & MCP.
Can Claude run AI scans or spend credits through the MCP server?
Claude cannot run live scans or add tracked prompts over MCP; those stay in the Measure app and Agent M. It can call estimate_scan_cost, which prices a scan at 1 credit per Perplexity answer, 2 per Google AI Overview and 3 per ChatGPT, Gemini or Claude answer, and shows your remaining balance. Industry-data and Google top 10 lookups are free from cache; a cache miss costs the same 1 to 45 credits as in the app.
Can Claude change anything in my Measure workspace?
Most of the 44 tools only read. A few write: Claude can update the shared project memory, add brand facts for the accuracy monitor, re-check AI accuracy verdicts, run and save a site audit, and save or delete reports. Every call is limited to the workspace you signed in with, and tools carry the MCP read-only and destructive hints.
How many MCP requests can I make on each Measure plan?
Tool calls are rate limited per workspace per minute: 60 on Starter, 120 on Pro, 300 on Advanced and 1,200 on Enterprise. A call over the limit is refused with a message to slow down and retry shortly. Page and site audits, technical checks and other heavier tools also share a separate, lower per-minute audit limit.
Know where your brand stands in AI answers.
Plans start at $89 a month. Add your brand, choose the prompts your buyers ask, and see who AI engines recommend.