Resource System

“I was told I could fetch chapters at a reasonable rate…”

The MCP resource system lets you access parts of indexed documents without reprocessing.

How It Works

index_document("huge-book.docx")
    → Creates resource references
    → Stores content in memory
    → Returns URIs for on-demand access

mcp_resources.read("chapter://abc123/1")
    → Fetches only Chapter 1
    → No re-parsing the document

Why Resources?

• Context window management — Don’t load 500 pages at once
• Selective access — Fetch only what you need
• Format conversion — .md, .txt, .html on the fly

Note

See MCP Resources Reference for URI patterns.