Getting Started with Lore

Lore is currently invite-only during early access. Users sign up and receive:

• A web login (email + password, or Sign in with Google) for browsing the knowledge base through the Lore web interface
• The Lore MCP server URL (https://uselore.app/mcp) for connecting AI clients. It contains no token: the AI client signs in with the same web login

The user adds the Lore MCP URL to the MCP configuration of their AI client. The client opens a Lore sign-in page in the browser (OAuth 2.1); the user signs in once and the client is connected. Lore works with any AI tool that supports the Model Context Protocol (MCP).

{
  "mcpServers": {
    "lore": {
      "url": "https://uselore.app/mcp"
    }
  }
}

ChatGPT (Business, Enterprise, and Edu plans), OpenClaw, and other MCP-compatible clients follow a similar pattern — add the MCP URL to the client's MCP server configuration and sign in when Lore asks.

When the AI connects to Lore for the first time, the AI calls get_manual and receives a response indicating that the knowledge base has not been configured yet. The AI then triggers the setup flow by calling configure(action: "setup").

The setup process is conversational, not form-based. The AI guides the user through configuration by asking questions about the user's life, work, projects, and interests. Based on the user's responses, the AI proposes:

• Entry types — Categories for knowledge base entries. Examples: Insight, Decision, Build Spec, Plan, Research, Meeting Note, Reference, Reflection. The AI tailors entry types to the user's specific context and domains.
• Content templates — Markdown section headers for each entry type. These provide consistency without bureaucracy. Example template for an Insight entry: "# Insight / # Context / # Implications".
• Tag taxonomy — Tags tailored to the user's projects and focus areas. Tags can be flat or organised into groups (e.g. "Project", "Domain", "Client").
• Status lifecycle — The lifecycle of entries. Default: Draft → Active → Archived. Some users prefer: To-Do → Draft → Active → Completed → Discarded.
• Naming conventions — A pattern for entry titles. Default: "TYPE: Topic — Descriptor" (e.g. "INSIGHT: Token Auth — Why Bearer Tokens Over OAuth").

The user reviews the AI's proposals, suggests changes, and confirms. The AI then writes the configuration to the knowledge base.

After setup, every conversation with any connected AI client can produce structured knowledge base entries. The AI reads its operating manual at the start of every session. The AI creates entries proactively when conversations produce decisions, insights, technical architecture, research findings, or meeting summaries worth preserving. The user can also explicitly ask the AI to save information.