claude-context-local

Claude Context Local is a semantic code search system that runs entirely on your machine using Google's EmbeddingGemma model. It integrates with Claude Code via MCP, allowing you to find code by meaning rather than string matching across Python, JavaScript, TypeScript, Go, Java, Rust, C, C++, and C# codebases. The tool uses intelligent AST-based and tree-sitter parsing to create semantically meaningful code chunks, then indexes them with FAISS for fast local similarity search. By keeping all embeddings and processing local, it eliminates API calls, reduces token usage in Claude prompts, and ensures your codebase never leaves your machine. This is especially valuable for teams managing large codebases who want to reduce API costs while maintaining code privacy.

["Identify the specific task or problem you need to solve in your codebase (e.g., 'Implement a caching layer for API responses').","Use the claude-context-local tool to search for relevant files, functions, or classes by specifying keywords or patterns (e.g., 'cache', 'memoization', 'Redis').","Review the tool's output to understand existing patterns, dependencies, or gaps in the codebase related to your task.","Refine your search if needed by adjusting keywords or adding constraints (e.g., 'Python', 'FastAPI', 'Django').","Use the insights to draft a solution, ensuring you reference specific files or functions for context.","Test your solution in a development environment and iterate based on feedback or errors."]