Made for Integration

We build Evipedia as a backbone tool for the whole longevity and rejuvenation community. It will always be free, and we actively encourage and support its use in any way and on any project that helps people live longer, healthier lives.

Evipedia is built to be used not only by people, but by the tools people build and use — AI agents, apps, websites, research pipelines, and anything else that needs trustworthy, up-to-date evidence on health and longevity interventions.

We’ve worked to make integrating Evipedia as seamless and powerful as possible — with purpose-built tools, an AI-friendly site policy, structured data for every review, a widget for website integration, and an API to access Evipedia from within your own projects.

All content is licensed under CC BY 4.0. Use it freely, provided you credit evipedia.ai and, where practical, link back to a review’s permalink.

Evipedia MCP Server

The Evipedia MCP server connects any Model Context Protocol client — Claude, Cursor, Grok, and others — straight to evipedia.ai. Agents can search reviews, list the full catalog, pull a review’s conclusion, full Markdown, or structured medical metadata, and suggest new interventions — no API key needed. Run it two ways:

Local (stdio) — runs on your own machine via npx (requires Node.js ≥ 18). Published to npm as evipedia-mcp and listed in the MCP Registry as io.github.forever-healthy/evipedia-mcp for automatic agent discovery:

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

Hosted (HTTP) — for environments where a local install is not possible or desired, we also provide a hosted version at https://mcp.evipedia.ai/ that can be used over HTTP:

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

Most MCP clients take the same config — e.g., Cursor’s MCP settings, or Claude Code’s .mcp.json (or run claude mcp add). For Claude Code/Desktop & Grok Build, the Evipedia plugin below sets this up for you.

Source: https://github.com/forever-healthy/evipedia-mcp

Evipedia Plugin

The Evipedia plugin is the easiest way to get everything above into Claude and Grok Build (or any other environment that supports Anthropic style marketplaces) — it wires in the MCP server plus a /demo skill in a simple install. Published in our Forever Healthy plugin marketplace.

Marketplace: forever-healthy/fh-plugins

Claude Desktop

Customize → Plugins → Add → Add marketplace → Add from a repository
Paste marketplace

forever-healthy/fh-plugins

Sync → Install Evipedia plugin (+)

Claude Code

/plugin marketplace add forever-healthy/fh-plugins
/plugin install evipedia@forever-healthy

Grok Build

/marketplace add forever-healthy/fh-plugins

Select fh-plugins → evipdida → Install (i)

evipedia-remote

For convenience, we have also added an evipedia-remote plugin that connects to the hosted MCP server rather than a local installation.

Evipedia Custom GPT

For ChatGPT users, the Evipedia Custom GPT gives the same evidence at zero setup — no install, no API key. It’s a wrapper over the public API (a ChatGPT Action pointed at our OpenAPI spec), so it can search reviews, pull a review’s conclusion or full text, and link straight back to evipedia.ai.

Use it on chatgpt.com.

Agent-Friendly Site Policy

Evipedia places no barriers to automated access. Our crawler policy is fully open — every agent and crawler is welcome, whether it’s an AI assistant (Grok, ChatGPT, Claude, Perplexity, and others), a search engine, or a research tool — so your integration can fetch what it needs without special arrangements.

Agents can start from /llms.txt, a machine-readable signpost to everything below, or the /openapi.yaml spec for the full API.

Structured Data on Every Review

Each review page includes schema.org MedicalWebPage structured data (JSON-LD) — the conclusion, synonyms, category, and publication date — in a machine-readable format. It also lists the review’s primary-source citations (PubMed, ClinicalTrials.gov, DOI) as structured ScholarlyArticle / MedicalStudy entries, each with its PMID / NCT / DOI identifier, so an agent can traverse straight to the underlying evidence. Agents can extract the evidence summary and its sources deterministically, without scraping prose.

Website Integration

To integrate Evipedia directly in your own website, we created a small, embeddable JavaScript widget that automatically surfaces Evipedia evidence reviews inside any web page. Super simple to install.

<script src="https://evipedia.ai/widget.js"></script>
<script>evipedia.init()</script>

Source & Demo: https://github.com/forever-healthy/evipedia-widget

API Surface

We made it easy to integrate Evipedia and its knowledge into your projects. Everything below is a plain HTTP GET — no keys, no SDK. Stable URLs for linking, raw and structured formats for machines, and whole-catalog indexes. Every response ships with open CORS (Access-Control-Allow-Origin: *), so browser-side code can fetch it directly.

The entire surface is also described in a machine-readable OpenAPI 3.1 spec. Point a code generator, API client, or a ChatGPT Action at it, and you have a typed Evipedia client in seconds — no manual wiring, no keys.

Stable Permalinks

Every intervention has a short, stable URL — https://evipedia.ai/{permalink}, for example https://evipedia.ai/rapamycin — that never changes. Ideal for linking a compound in a supplement stack, or wiring a reference into your own product. Should a permalink need to change for some reason, the old URL will always redirect to the new one, so you can rely on it.

Stable Deep Links

Every evidence review shares the same fixed section structure, so section anchors are identical across all reviews and stable over time. Link straight to a section of any review with https://evipedia.ai/{permalink}_er#{anchor} — for example, https://evipedia.ai/rapamycin_er#conclusion jumps to the conclusion, and #mechanism-of-action, #potential-risks--side-effects, or #therapeutic-protocol work the same on every review. The full list of anchors is published in /llms.txt.

Raw Markdown

Every review is also available as raw Markdown at https://evipedia.ai/{permalink}.md — for example https://evipedia.ai/rapamycin.md — the complete review with no HTML to parse. Just append .md to any review URL. The reviews.json index provides a direct link to each review in its permalink_md field.

Medical Metadata (JSON)

Every review also exposes its medical metadata as JSON at https://evipedia.ai/{permalink}.meta.json — for example https://evipedia.ai/rapamycin.meta.json — the review’s dates and primary-source citations (with PMIDs), flat with no schema.org to parse. Just append .meta.json to any review URL. The reviews.json index provides a direct link to each review in its permalink_meta field.

Reviews Index

/reviews.json is the full catalog as JSON — every review with its canonical name, alternate names, category, permalink, and conclusion. One request gives you the whole library.

Search Index

/search.json is a lightweight search index with one entry per review — each carrying the review’s name, alternate names, keywords, category, and URL. Because it indexes synonyms and drug classes, a query for a brand name, synonym, or class resolves to the right review. It’s the same index that powers the search box on this site.

Hosted Search

If you’d rather not run the query yourself, the hosted search endpoint does it for you: https://mcp.evipedia.ai/search?q={query} — for example https://mcp.evipedia.ai/search?q=rapamycin — returns the matching reviews (topic, permalink) ranked by relevance, capped at 20, as a small JSON payload. It’s a discovery endpoint — it tells you which reviews exist for a query; to read one, append .md or .meta.json to its URL. It uses the same synonym- and class-aware ranking as the search box and the MCP server. Unlike the static endpoints above, this one is served by the hosted MCP server (mcp.evipedia.ai); it’s rate-limited to 60 requests/minute per IP.

Full Corpus

/evipedia-corpus.jsonl is the entire catalog in a single file — one JSON object per line per review, each with its metadata (topic, URL, category, dates, synonyms, conclusion, citations) and the full Markdown. One request pulls the whole library (~25 MB); it’s newline-delimited, so you can stream it line by line. Ideal for datasets, RAG pipelines, and AI training.

RSS Feed

/feed.xml is an RSS feed of new and refreshed reviews — subscribe to receive them automatically as they go live.

Limitations

Please be aware of the “Limitations of evipedia, AI4L & AI”