MCP Directory

How to add LLM Context to Claude Desktop

Smart context management for LLM workflows — rule-based file selection shared via MCP or clipboard. Paste the config into ~/Library/Application Support/Claude/claude_desktop_config.json and restart Claude Desktop.

Last updated June 14, 2026 · 301 · stdio · no auth

Claude Desktop config for LLM Context

uv tool install "llm-context>=0.6.0"
{
  "mcpServers": {
    "llm-context": {
      "command": "uvx",
      "args": [
        "--from",
        "llm-context",
        "lc-mcp"
      ]
    }
  }
}

Requires `uv` (the Python package runner). Install it from https://docs.astral.sh/uv/ if `uvx` is not found.

Setup steps

  1. 1Open Claude Desktop → Settings → Developer → Edit Config (this opens ~/Library/Application Support/Claude/claude_desktop_config.json).
  2. 2Paste the LLM Context config below under the top-level "mcpServers" key.
  3. 3Fill in any placeholder secrets (API keys, paths) in the snippet.
  4. 4Save the file, then fully quit and reopen Claude Desktop.
  5. 5Open a chat and confirm LLM Context's tools appear under the 🔌 tools menu.

Before you start

  • uv / uvx installed (used to install and run the lc-* commands and the MCP server)
  • A project initialized with lc-init (creates .llm-context/)

What LLM Context can do in Claude Desktop

lc_outlines

Generate excerpted context (code structure outlines) from the current rule.

lc_preview

Validate a rule's effectiveness — check file selection and context size before use.

lc_missing

Fetch specific files or implementations on demand (by param_type and data, with a timestamp).

Security

The server exposes filesystem read access (file selection, outlines, and on-demand file fetching) scoped to the project root path you provide. It runs locally over stdio with no authentication; only grant it access to projects you intend the AI to read.

LLM Context + Claude Desktop FAQ

Where is the Claude Desktop config file?

Claude Desktop reads MCP servers from ~/Library/Application Support/Claude/claude_desktop_config.json. Paste the LLM Context config there under the "mcpServers" key and restart the client.

Is LLM Context safe to use with Claude Desktop?

The server exposes filesystem read access (file selection, outlines, and on-demand file fetching) scoped to the project root path you provide. It runs locally over stdio with no authentication; only grant it access to projects you intend the AI to read.

Do I need the MCP server, or can I just use the clipboard?

Both work. The clipboard workflow (lc-select / lc-context) requires no server. The MCP integration (lc-mcp) is recommended so the AI can fetch additional files during a conversation without manual copying.

How do I add it to Claude Desktop?

Add an mcpServers entry to claude_desktop_config.json with command "uvx" and args ["--from", "llm-context", "lc-mcp"], then restart Claude Desktop.

What are rules?

Rules are YAML+Markdown files describing what context to provide for a task. They come in five categories (prm-, flt-, ins-, sty-, exc-) and can be composed from simpler reusable rules.

View repo Full LLM Context page