Knowledge layer for AI

Give every AI tool
your full context

Your knowledge is scattered across tools. Lore brings it together — so Claude Code, ChatGPT, Cursor, and any MCP client can all work with the same understanding.

npm install -g @mishkinf/lore

Context that doesn't lose the source

Memory tools store summaries. Lore preserves your original documents and lets any tool cite them.

Without Lore

“Users want faster exports”

With Lore

“In the Jan 15 interview, Sarah said ‘The export takes forever, I’ve lost work twice this week’”

lore browse
Lore TUI — browse documents, search, and research from the terminal

Browse and search your knowledge base from the terminal.

Everything in one place

Hybrid Search

Find by meaning or exact terms.

Citations

Every result links back to the original source.

MCP Tools

9 tools for Claude Desktop, Claude Code, and more.

Multi-Machine Sync

Same knowledge base everywhere. No duplicates.

Agentic Research

AI explores, cross-references, and synthesizes.

Universal Formats

Markdown, PDF, images, CSV, HTML, and more.

Point. Sync. Search.

Sync Sources Point at directories with your documents
Index Claude extracts metadata, generates embeddings
Search & Cite Find what you need with full source citations

Up and running in minutes

Requires an OpenAI API key (embeddings) and Anthropic API key (research).

1

Install

npm install -g @mishkinf/lore
2

Setup

Configure API keys and sign in:

lore setup
3

Add Sources

Point at a folder of documents:

lore sync add
4

Sync & Search

lore sync
lore search "user pain points"
lore research "What should we prioritize?"

Use with Claude Desktop & Claude Code

Add Lore as an MCP server. API keys from lore setup are used automatically.

Add to .mcp.json in your project or ~/.claude.json globally:

{
  "mcpServers": {
    "lore": {
      "command": "lore",
      "args": ["mcp"]
    }
  }
}

Settings → Developer → Edit Config. Include API keys since Desktop doesn't inherit your shell:

{
  "mcpServers": {
    "lore": {
      "command": "lore",
      "args": ["mcp"],
      "env": {
        "OPENAI_API_KEY": "your-key",
        "ANTHROPIC_API_KEY": "your-key"
      }
    }
  }
}

Add wherever your client expects MCP server definitions:

{
  "command": "lore",
  "args": ["mcp"],
  "env": {
    "OPENAI_API_KEY": "your-key",
    "ANTHROPIC_API_KEY": "your-key"
  }
}

Available MCP Tools

search Find by meaning or exact terms
get_source Get full document content
list_sources Browse by project or type
list_projects Overview of all projects
retain Save an insight or decision
ingest Add a document directly
sync Sync sources
archive_project Archive a completed project
research AI-powered deep research

Command overview

CommandDescription
lore setupGuided configuration wizard
lore loginSign in with email OTP
lore syncSync all configured sources
lore sync --dry-runPreview what would sync
lore sync addAdd a source directory
lore sync listList configured sources
lore search <query>Hybrid search
lore research <query>AI-powered deep research
lore browseInteractive TUI browser
lore docs listList all documents
lore docs get <id>View a document
lore projectsList projects
lore projects archive <name>Archive a project
lore mcpStart MCP server

Common questions

What's the difference between Lore and a memory system?

Memory systems store processed facts without attribution. Lore preserves original documents so you can cite exactly what was said, by whom, and when.

What does sync cost?

Only new files cost anything to process. Re-syncing existing files is free.

Can multiple people share a knowledge base?

Not yet. Each user's data is isolated. Team sharing is planned.

What file formats are supported?

Markdown, JSONL, JSON, plain text, CSV, HTML, XML, PDF, and images. Claude extracts metadata automatically.

Do I need to set up any infrastructure?

No. Backend is fully hosted. Just install, log in, and bring your own API keys.

How does multi-machine sync work?

Sync the same folders from multiple machines. Lore skips anything already indexed.