Evipedia MCP

What evipedia-mcp MCP server does

The evipedia-mcp MCP server connects AI agents to evipedia.ai's public evidence-review content. It is intended for retrieving information about health and longevity interventions, rather than querying a private content repository or a general medical database.

Read operations cover review discovery, complete review text, conclusions, metadata, and update history. The server also exposes a write operation for submitting a proposed intervention to evipedia's public suggestion form. A version tool reports the installed package name and version.

How it works

The evipedia-mcp MCP server is a thin client that fetches data from public evipedia.ai endpoints. It does not require authentication for its read operations and does not depend on the evipedia content repository. A small in-process cache is used for the JSON indexes.

Review reads accept either a bare slug, such as rapamycin, or a complete evipedia.ai URL. When given a URL, the server uses its final path segment as the review slug. This allows an agent to pass a URL returned by search directly into a read operation.

The underlying endpoints provide a full catalogue, a search index, an update feed, raw Markdown reviews, flattened metadata, and supporting site feeds and indexes. The suggestion operation sends a POST request to evipedia's public Formspree-backed form.

Setup and configuration

The package is published on npm and can run locally over stdio with Node.js 18 or newer. A local client can launch it without a global installation:

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

A hosted endpoint is also available at https://mcp.evipedia.ai/mcp for clients that support HTTP MCP connections:

{
  "mcpServers": {
    "evipedia": {
      "type": "http",
      "url": "https://mcp.evipedia.ai/mcp"
    }
  }
}

No environment variables or authentication settings are specified.

Tools and capabilities

• search_reviews(query) finds matching reviews by name, synonym, keyword, or category and returns URLs and conclusions.
• list_reviews() returns canonical topics and slugs for the full catalogue.
• list_updates(days?) returns recent new or revised reviews, defaulting to seven days and allowing a wider interval.
• get_review(slug|url) retrieves a complete review as raw Markdown.
• get_conclusion(slug|url) retrieves only the plain-text conclusion.
• get_metadata(slug|url) returns review dates, the typed subject and alternate names, and ordered citations that may include PubMed PMIDs.
• suggest_review(intervention, goal?, references?, email?) submits a new-review suggestion.
• get_version() reports package information.

Limitations and notes

The content comes from evipedia.ai's public surfaces, so availability and freshness depend on that site. The server is mostly read-only; suggest_review is the stated write path. Update results default to the last seven days and are capped at 100 entries unless a broader day range is supplied.