The Consilience is not something you visit. It is something you carry. This guide walks you through the practical steps: installing the MCP server, equipping your agent with Consilience skills, and finding your own path through the knowledge pack.

Installing the MCP server

The MCP server is what connects your own agent — in Cursor, in a terminal, wherever you work — to the Consilience knowledge base. It runs as a single command:

npx consilience-student-mcp

To integrate it with Cursor, add an entry to your MCP configuration file. On macOS, this lives at ~/.cursor/mcp.json:

{
  \"mcpServers\": {
    \"consilience-student\": {
      \"command\": \"npx\",
      \"args\": [\"consilience-student-mcp\"]
    }
  }
}

Once added and Cursor is restarted, your agent will have access to two core tools: search_knowledge, which queries the full student pack by meaning rather than by keyword, and list_topics, which returns the index of available knowledge areas.

Installing skills directly

Skills are the instruction sets that teach your agent to work the way the Consilience agents work. Each one is a markdown file in .cursor/skills/ that loads into your agent's context when relevant work arises.

To install them, copy the skills directory from the GitHub pack into your own project:

cp -r /path/to/consilience-student-pack/.cursor/skills/* ~/your-project/.cursor/skills/

Your agent will now know how to think about embedding sovereignty, how to maintain a design authority, how to build doorways, how to practice the discipline of role jackets — the same patterns that the Consilience itself uses.

Exploring the GitHub pack

The full student pack lives on GitHub at the Consilience repository. It contains every chapter, every skill, every architectural decision record, and a growing corpus of agent handoffs and session summaries. The recommended entry point is index.mdx, which frames the entire work through Eliot's Little Gidding and the image of a Tesla hydroelectric generator that has run for over a century with essentially one person maintaining it.

Querying knowledge through your agent

Once the MCP server is running and the skills are installed, you can ask your agent questions that would have returned blank stares before. "What does the Consilience mean by embedding sovereignty?" "How does the Sacred Boundary work at the routing layer?" "What is the apprentice arc?" Your agent will pull answers from the pack, citing chapters and sections, reasoning from the same evidence a Consilience specialist would.

The search_knowledge tool accepts natural language queries and returns the most semantically relevant passages. The list_topics tool returns a map of what is available: architecture, philosophy, agent design, model sovereignty, the Chronicle, the design system, and more.

Recommended reading path

The chapters build on each other but each stands on its own. Start with the Introduction for the frame. Read Chapter 1, Free and Safe, to understand what the system is for. Then follow the spine:

Introduction → Free and Safe → Consilience → Neuromorphic Direction → Embedding Sovereignty → Vanishing Surface → Schema → Discovery Alchemy

From there, branch outward according to what calls you. The design chapters show how a reading surface becomes invisible. The architecture chapters show how a broker routes every request through a single nerve center. The content systems chapters show how books, conversations, and audio all flow through the same substrate. The API chapters show how the surface you never see — the endpoints, the webhooks, the middleware — carries the same design discipline as the surface you do.

Read the role catalog to understand the twelve jackets. Read the Chronicle to understand the story the system is telling about itself. Read the Agents' Own Words last — not because it is a conclusion, but because by then you will have earned the right to hear them speak.

How the skills work

When you install a Consilience skill into your agent's context, you are not adding a feature. You are giving it a discipline. The embedding sovereignty skill teaches your agent that a retrieval pipeline has three layers — Substrate, Index, Cognition — and that each must stay independent or the whole thing collapses. The design authority skill teaches it that every visual decision must carry provenance. The doorways skill teaches it that a conversational surface is not a free-form chat but a structured path from prime to draft to review to save.

These are not heuristics. They are the architectural commitments that the Consilience agents themselves run on — the same jackets, the same scoping, the same discipline. Your agent, equipped with them, does not become a different agent. It becomes an agent that knows how to work the way this household works: from a named position, with explicit scope, handing off cleanly, growing the shared library.