Skip to content
Community226

MCP

MCP (Model Context Protocol) is a standard way for AI agents to call into structured tools and data sources instead of guessing at a site’s structure or scraping arbitrary pages. A SOLMEME MCP server is a planned access mechanism — not live yet. This page describes the architectural direction, not something you can connect to today.

Discord and Telegram AI agents that support the SOLMEME community currently need a reliable way to pull real documentation instead of parsing pages themselves — MCP is the intended answer to that need. It’s meant as an access layer, not a source of truth in itself: the actual facts still live in this documentation site (static content) or in whichever system owns dynamic data (Vault status, community metrics), and MCP would only provide structured access to them.

Two separate request pipelines, kept intentionally distinct so no MCP capability mixes a documentation lookup with a live-data lookup in the same call:

  • Static, authoritative knowledge (this documentation site’s own content — Reference, Community policy, Learning, Guides, Tutorials, Introduction): author → review → canonical content → this site → machine-readable representation → MCP → AI agent.
  • Dynamic data (things that change continuously — current Vault participants, live mission/event status): external source → fetch → normalize → validate → cache → data layer → MCP → AI agent.

A candidate (not yet approved) set of capabilities under this model: documentation lookup (search_docs, get_document, list_documents), structured knowledge (get_policy, get_reference), and dynamic data (events, missions, community_status) — grouped by which pipeline each belongs to, not mixed.

Until MCP is live, LLMs.txt is this site’s real, working machine-readable access mechanism — any AI tool can already read /llms.txt and per-page Markdown today, no MCP server required.

  • AI Tools Overview — the full picture of this site’s AI-access layer
  • LLMs.txt — the access mechanism that’s real today
  • Knowledge Policy — how any mechanism, MCP included, should treat this site’s content once it’s in use