llm-context.py

Overview

LLM Context is a Python-based tool that addresses the challenge of getting the right context into LLM conversations. It provides focused, task-specific project context through composable rules that allow intelligent file selection and filtering. The tool works both for humans using chat interfaces and AI agents with CLI access, offering features like smart code outlining, context validation, and MCP integration for seamless file access during conversations.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• lc_outlineslc_outlines(root_path, rule_name)

  Generate excerpted context from a specific rule in the project

  Try:Show me the code structure for the authentication moduleTry:Generate context for the OAuth implementation using the tmp-prm-oauth ruleTry:Get outlines for the payment processing files
• lc_previewlc_preview(root_path, rule_name)

  Validate a rule's effectiveness by showing which files it would include

  Try:Check which files the tmp-prm-auth rule would selectTry:Preview the files selected for the OAuth implementation ruleTry:Validate the payment processing rule before using it
• lc_missinglc_missing(root_path, param_type, data, timestamp)

  Fetch specific files or implementations on demand

  Try:Get the authentication implementation from /src/auth.pyTry:Fetch the OAuth test filesTry:Retrieve the payment processing code from /proj/src/payment.py

Comparable tools

Installation

# Install as a tool
uv tool install "llm-context>=0.6.0"

# MCP Integration for Claude Desktop
Add to Claude Desktop config (~//Library/Application Support/Claude/claude_desktop_config.json):
```jsonc
{
  "mcpServers": {
    "llm-context": {
      "command": "uvx",
      "args": ["--from", "llm-context", "lc-mcp"]
    }
  }
}

cd your-project
lc-init

FAQ

How does llm-context select files?

It uses YAML+Markdown rules that describe what context to provide for a specific task. Rules can include/exclude files based on patterns, define excerpting strategies, and combine multiple rule components.

What MCP tools are provided?

lc_outlines - Generate excerpted context from current rule, lc_preview - Validate rule effectiveness before use, lc_missing - Fetch specific files/implementations on demand.

Compare llm-context.py with