Machine Access

For AI Agents

Mystery Schools AI exposes its knowledge graph, corpus, and series architecture through an MCP server and a suite of machine-readable text endpoints. Eight tools. 4,691 embedded passages. 713 KB entries. No scraping required.

119Concepts
130Figures
122Connected sources
4,691Corpus passages
8MCP tools

MCP Server

Connect any MCP-compatible client to query the knowledge graph, search the corpus, and retrieve structured episode and series data.

MCP access is approved by request in phase 1. It is a separate developer path — not bundled into standard membership. Approved accounts receive a user-bound API key. Send it as x-api-key or as a Bearer token.

Contact the project to request developer access.

Transport endpoint

https://mystery-schools.vercel.app/api/mcp/sse

Cursor — .cursor/mcp.json

{
  "mcpServers": {
    "mystery-schools": {
      "url": "https://mystery-schools.vercel.app/api/mcp/sse",
      "headers": { "x-api-key": "YOUR_KEY" }
    }
  }
}

Claude Desktop — claude_desktop_config.json

{
  "mcpServers": {
    "mystery-schools": {
      "type": "sse",
      "url": "https://mystery-schools.vercel.app/api/mcp/sse",
      "headers": { "x-api-key": "YOUR_KEY" }
    }
  }
}

Tools

Eight tools registered on the server. Each returns structured text for agent consumption.

search_corpus

Semantic corpus search

4,691 embedded passages from 154 texts — classical philosophy, Hermetic literature, Vedic scripture, initiatic traditions, and critical theory.

params: query, topK (max 20)

get_concept

Concept lookup

Retrieve any of 83+ concepts by name or ID. Initiation, Katabasis, Epopteia, Participation, Gnosis, Theurgy, Apophatic, Mundus Imaginalis, and more.

params: name or id

get_figure

Figure biography

Intellectual biographies of 113 thinkers: Eliade, Barfield, Gebser, Plotinus, Guénon, Corbin, Burkert, Steiner, and others. Dates, domain, key works, KB relations.

params: name or id

get_sources

Library sources

Source texts connected to the knowledge graph — 346 books with bibliographic data. Filter by category.

params: category (optional)

get_knowledge_graph

Full knowledge graph

All nodes (concepts, figures, books, timeline events) and all named relations between them.

params: none

browse_imagery

Imagery corpus

109 curated images: alchemical manuscripts, Hermetic illustrations, classical archaeology, portraits, Neoplatonic diagrams. Filter by category or search by tag.

params: category, search (both optional)

get_series

Series architecture

Full series architecture across 7 tracks, 25 series, ~231 episodes. Filter by track.

params: track (optional)

get_episode

Episode metadata

Retrieve episode status, featured concepts and figures, primary sources, and media availability. Lookup by ID or title, or filter by status and track.

params: id, status, track (all optional)

Text Endpoints

Public, unauthenticated, cached. For crawlers and agents that prefer plain text over MCP.

/llms.txt

Project overview and index of all machine-accessible surfaces. Start here.

/llms-full.txt

Full public content dump — concepts, figures, sources, graph, series, imagery in one file.

/series.md

Complete episode listing across all tracks and series.

/concepts.md

All concepts with definitions, traditions, and cross-references.

/figures.md

All figures with biographies, dates, and key works.

/sources.md

All connected library sources with bibliographic data.

/explore.md

Full knowledge graph as plain text — all nodes and edges.

Knowledge Base

Three entity types with named relations between them. Every episode references KB entries; every KB entry cross-references related entries.

Concepts119Philosophical and initiatic concepts with tradition tags and episode appearancesBrowse →Figures130Thinkers, initiates, and scholars with intellectual biographiesBrowse →Sources122Library texts connected to the graph with bibliographic metadataBrowse →
Interactive knowledge graph →

Access tiers. Text endpoints and the public knowledge base are open to everyone, no key required. Semantic search over the knowledge base requires a free sign-in. MCP is a separate developer surface available by request in phase 1 — approved accounts receive user-bound keys. Standard membership does not automatically include MCP access.

Rate limits. The MCP server allows 120 requests per hour. Text endpoints are public and cached at the edge.

Copyright. The corpus contains copyrighted material used under research and commentary principles. MCP tools return interpretive summaries and selected passages, not verbatim reproductions.

Developer access. To request an approved developer key, see the project log.

0:00
0:00