AI Access – MCP Server | Ladislav Hejdánek Archive

AI Access (MCP Server)

This website runs a server based on the Model Context Protocol (MCP). It allows language models (LLMs) direct, structured access to the Ladislav Hejdánek text archive.

Technical Parameters

Protocol	MCP (Model Context Protocol)
Connection type	SSE (Server-Sent Events)
Endpoint	https://www.hejdanek.eu/mcp
Methods	Semantic search (vectors), document retrieval by ID, metadata

Client Configuration

Use the following settings to connect the archive to your AI assistant:

Cursor

In settings (Settings → Features → MCP Servers) add a new server:

Type: SSE
URL: https://www.hejdanek.eu/mcp

Claude Desktop

Add the following configuration to your mcp.json file:

{
  "mcpServers": {
    "hejdanek": {
      "type": "sse",
      "url": "https://www.hejdanek.eu/mcp"
    }
  }
}

Available Tools

Once connected, the AI agent has access to the following functions:

`SearchTexts`	Semantic search across the corpus based on meaning similarity (powered by PostgreSQL pgvector). Supports optional filters: document type, year of origin (from/to), author, language, published texts only.
`GetDocumentText`	Retrieve the full OCR text of a document by its ID. The ID can be obtained from `SearchTexts` results.
`GetFilterOptions`	Returns the list of valid values for the filter parameters of `SearchTexts` (document types, languages). Recommended to call before using filters.

Information for Agents and Automated Systems

A discovery configuration file is available at:
https://www.hejdanek.eu/.well-known/mcp.json

Document metadata is indexed in JSON-LD format. Source texts are stored in UTF-8 encoding.