Ask Claude about your AI visibility

The server

URL

https://saymeter.com/api/mcp

Transport

Streamable HTTP (stateless)

Authentication

Header Authorization: Bearer smk_…, an API key from Integrations in your workspace

Limit

60 requests a minute per key, shared with the REST API

Make a separate key for each assistant, so you can revoke one without breaking the others. Anyone with the key can read the whole workspace.

Tools

• list_locations: Every tracked location in this workspace with its id and latest overall reading.
• get_standing: The latest reading for one location: its share of AI answers and each competitor's, overall and per engine, and the change since the previous like-with-like reading.
• get_trend: A location's overall share in each recent reading, oldest first, labelled by kind of reading.
• list_fixes: The location's fix cards: what to change on the site or listings, by impact.
• get_answers: What the AI engines actually said in a location's latest reading, optionally for one question and/or one engine, with who each answer named.
• compare_competitors: Each competitor's share in the latest reading against the previous like-with-like reading, and whether any clearly overtook the location.

Answer text from the engines is passed to your assistant marked as third-party data. Assistants can still be misled by text they read, so check anything surprising against the answer in the app.

Claude (web and desktop apps)

Custom connectors take a fixed request header, which is how the API key is sent. On a Pro or Max plan:

1. Open Customize, then Connectors, and add a custom connector.
2. Name it Saymeter and paste the URL https://saymeter.com/api/mcp.
3. For authentication choose no sign-in.
4. Under Request headers, add Authorization with the value Bearer smk_your_key, then add the connector.

On Team and Enterprise plans an owner adds it once under Organization settings → Connectors, and every member who connects shares that key, so use a key made for the purpose. Free accounts can add one custom connector.

Steps from Anthropic's custom connectors guide, checked 18 October 2026. Menu labels change; the guide is the authority.

Claude Code

claude mcp add --transport http saymeter https://saymeter.com/api/mcp \
  --header "Authorization: Bearer smk_your_key"

Or commit it to a project's .mcp.json and keep the key in an environment variable:

{
  "mcpServers": {
    "saymeter": {
      "type": "http",
      "url": "https://saymeter.com/api/mcp",
      "headers": { "Authorization": "Bearer ${SAYMETER_API_KEY}" }
    }
  }
}

Syntax from the Claude Code MCP docs, checked 18 October 2026.

Claude API

The Messages API's MCP connector sends the key as a bearer token:

// Messages API, with the header  anthropic-beta: mcp-client-2025-11-20
{
  "model": "…",
  "max_tokens": 1024,
  "messages": [{ "role": "user", "content": "How did my locations move this week?" }],
  "mcp_servers": [
    { "type": "url", "url": "https://saymeter.com/api/mcp", "name": "saymeter", "authorization_token": "smk_your_key" }
  ],
  "tools": [{ "type": "mcp_toolset", "mcp_server_name": "saymeter" }]
}

Shape and beta header from Anthropic's MCP connector docs, checked 18 October 2026.

ChatGPT and OpenAI

The ChatGPT app can't use this server yet. Its custom connectors (developer mode) support OAuth or no authentication, and OpenAI states they can't present custom API keys (OpenAI's authentication docs). A server that answered without a key would expose your data, so we don't offer one.

OpenAI's Responses API can send the key in a header, so you can use the server from your own OpenAI-based tools:

// Responses API
{
  "model": "…",
  "input": "Which of my clients lost ground to a competitor this week?",
  "tools": [{
    "type": "mcp",
    "server_label": "saymeter",
    "server_url": "https://saymeter.com/api/mcp",
    "headers": { "Authorization": "Bearer smk_your_key" },
    "require_approval": "never"
  }]
}

From OpenAI's Responses API reference, checked 18 October 2026.