LLM Pulse MCP Server for Claude

Content

LLM Pulse MCP Server connects Claude and other MCP-capable clients to LLM Pulse, an AI visibility analytics platform for monitoring how brands appear in AI search and LLM-generated answers. Teams use it to inspect brand mentions, citations, sentiment, competitor share of voice, weighted visibility, prompt coverage, source domains, recommendations, GEO Writer tasks, Search Console metrics, and AI traffic data from a conversational workflow.

The hosted endpoint is:

https://api.llmpulse.ai/api/v1/mcp

Supported clients can connect through remote Streamable HTTP and authenticate with OAuth. LLM Pulse also documents an API-key fallback for headless integrations on Scale and Enterprise plans, plus a small public wrapper repository for local stdio testing.

Source Review

• https://llmpulse.ai/features/mcp
• https://llmpulse.ai/api-docs
• https://github.com/LLM-Pulse/llmpulse-mcp
• https://raw.githubusercontent.com/LLM-Pulse/llmpulse-mcp/main/README.md
• https://raw.githubusercontent.com/LLM-Pulse/llmpulse-mcp/main/server.json
• https://raw.githubusercontent.com/LLM-Pulse/llmpulse-mcp/main/package.json
• https://raw.githubusercontent.com/LLM-Pulse/llmpulse-mcp/main/src/index.js

These sources were reviewed on 2026-07-06. Prefer the live product docs, API docs, public metadata repository, server manifest, package metadata, and wrapper source for current setup and operating details.

Features

• Remote Streamable HTTP MCP endpoint for LLM Pulse account data.
• OAuth connection path for supported clients, with token refresh handled by the client flow.
• Headless API-key fallback for Scale and Enterprise integrations.
• Project, competitor, tag, model, locale, prompt, execution, citation, source, mention, sentiment, and answer discovery tools.
• Timeseries, summary, share of voice, prompt summary, top source, AI model, and AI traffic analytics where plan access allows.
• Recommendation and GEO Writer workflows for turning visibility data into content and optimization actions.
• Write-capable workflows for adding prompts, competitors, tags, annotations, technical GEO reports, and related account actions when scopes permit.
• Public local wrapper that can expose the hosted tool set through stdio when LLMPULSE_API_KEY is configured.

Use Cases

• Ask Claude for this month's AI visibility trend across ChatGPT, Perplexity, Gemini, Google AI surfaces, and other tracked models.
• Compare brand share of voice against tracked competitors for a project, locale, model, or prompt collection.
• Pull citations and source domains behind positive, neutral, or negative AI answers.
• Investigate weak prompts, low weighted visibility, missing citations, and sentiment changes without exporting CSV files.
• Draft a GEO Writer brief or recommendation from live LLM Pulse metrics.
• Review AI bot traffic and Search Console context alongside AI visibility data where the connected account has access.
• Let an internal agent monitor brand visibility and propose prompt, competitor, tag, annotation, or content updates for human review.

Installation

Claude Code

Add the hosted MCP endpoint:

claude mcp add --transport http llm-pulse https://api.llmpulse.ai/api/v1/mcp

Open the MCP connection and complete the LLM Pulse OAuth flow when prompted by your client.

Claude, ChatGPT, Gemini, Cursor, Windsurf, Zed, and custom clients

Add the same URL as a custom MCP connector or remote HTTP MCP server:

https://api.llmpulse.ai/api/v1/mcp

Use OAuth in clients that support it. For headless automations on Scale and Enterprise, use the documented bearer-token fallback:

npx mcp-remote https://api.llmpulse.ai/api/v1/mcp --header "Authorization: Bearer YOUR_API_KEY"

Local stdio wrapper

Clone the public wrapper repository and run it with an API key:

npm install
LLMPULSE_API_KEY=llmpulse_your_key_here npm start

Without LLMPULSE_API_KEY, the wrapper exposes a setup/status tool so registries and users can verify that the server starts before connecting real account data.

Configuration

Remote clients can use the hosted endpoint directly:

{
  "mcpServers": {
    "llm-pulse": {
      "url": "https://api.llmpulse.ai/api/v1/mcp",
      "type": "http"
    }
  }
}

Headless bearer-token setups should keep the API key in client secrets or environment configuration:

{
  "mcpServers": {
    "llm-pulse": {
      "url": "https://api.llmpulse.ai/api/v1/mcp",
      "type": "http",
      "headers": {
        "Authorization": "Bearer llmpulse_your_key_here"
      }
    }
  }
}

Example Prompts

Visibility trend

Show my brand visibility trend for the last 30 days and explain the biggest model-level changes.

Competitor share of voice

Compare our share of voice against our top competitors for the current project this quarter.

Citation investigation

Which domains cite us most often in AI answers, and which high-value prompts still have weak citation coverage?

GEO action plan

Find three prompts where we are visible but poorly positioned, then draft a GEO Writer brief to improve the weakest cluster.

Safety and Privacy

Treat LLM Pulse MCP as access to live account analytics. Read calls can return tracked prompts, brand names, competitor names, AI answer text, citation URLs, source domains, sentiment labels, Search Console metrics, AI traffic data, and recommendations. Only connect clients approved to process that data, and avoid pasting OAuth tokens or API keys into chat.

Write scopes can modify account objects or start workflows. Review any AI-proposed prompt, competitor, tag, annotation, content, or technical audit change before approving it, especially inside shared customer, agency, or white-label workspaces.

Troubleshooting

OAuth connection fails

Confirm that your MCP client supports remote Streamable HTTP and OAuth for custom connectors. If the client does not support OAuth yet, use the documented API-key fallback only where your plan and security policy allow it.

Tools are missing

LLM Pulse filters the MCP tool list by plan, scopes, and team-member permissions. Re-authenticate with the correct account, check the user's project access, and confirm that the requested feature is available on the current plan.

Headless wrapper only shows status

Set LLMPULSE_API_KEY before starting the wrapper. Without an API key, the wrapper intentionally exposes only the setup/status tool.

• Source repositorygithub.com 2026-07-06T14:41:13+00:00
• Documentationllmpulse.ai
• Websitellmpulse.ai
• Submitted by estevecastells2026-07-06