Pulse MCP

What pulse-mcp MCP server does

The pulse-mcp MCP server gives an MCP-compatible agent access to public social-media metrics through the Pulse API. An agent submits a post or profile URL and receives structured fields instead of parsing a web page. Post results can include the platform, views, likes, comments, shares, quotes, bookmarks, publication time, title, author, and thumbnail. Profile results include the handle, display name, follower and following counts, post count, verification state, and avatar.

The service covers public data from YouTube, X/Twitter, TikTok, Instagram, Bluesky, and Mastodon. Threads supports profile metrics, while its post metrics require login. LinkedIn post and profile data are login-walled. Platforms do not expose identical fields; for example, Bluesky has no view count, and Mastodon reports likes, boosts, and replies.

How it works

The MCP process forwards URL-based requests to Pulse, which reads each platform’s public pages or public APIs from a residential IP. That approach is intended to reflect browser-visible data. Deleted or private content is reported as unavailable, and login-restricted content is identified as requiring login rather than being represented as zero-valued metrics.

Single lookups handle one URL at a time. Batch operations accept multiple URLs, preserve profile result order, and allow individual failures without discarding successful results. The history operation returns snapshots collected during fresh fetches. For posts, points contain a timestamp and engagement counts; for profiles, points contain a timestamp, follower count, and post count. An optional since value can be supplied for a history request.

Setup and configuration

Add the server to an MCP client configuration using the package runner:

{
  "mcpServers": {
    "pulse": {
      "command": "npx",
      "args": ["-y", "pulse-mcp"]
    }
  }
}

No account, API key, or wallet is needed. By default, the process uses https://pulse.walls.sh as its backend. Set PULSE_API_URL to use a different backend endpoint. The repository is released under the MIT license.

Tools and capabilities

The pulse-mcp MCP server exposes five tools:

• metrics retrieves post metrics from one URL. Short links such as t.co and vm.tiktok.com are supported.
• metrics_batch retrieves the same post-oriented fields for multiple URLs. The input may contain a mixture of post and profile URLs, and partial failures do not fail the whole request.
• history returns the recorded growth curve for one post or profile, with an optional starting point for the requested delta.
• profile fetches account-level metrics for one profile URL across supported platforms.
• profile_batch fetches multiple profiles, supports up to 50 URLs, preserves input order, and tolerates partial failures.

These operations can support tasks such as comparing creator follower counts, checking how a public video performed, or tracking changes across repeated fetches.

Limitations and notes

Results depend on what each platform exposes publicly. YouTube returns views and likes, while X can also provide comments, reposts, quotes, and bookmarks, subject to the stated coverage limits. Instagram post metrics use public guest access. Threads post metrics and LinkedIn data may return a login-required status. Deleted or private posts may return a content-unavailable status.

Profile batch requests are limited to 50 URLs. Historical data is based on snapshots taken when fresh fetches occur, so the tool does not imply a continuously sampled timeline.