CodeWeaver

概要

CodeWeaver is an extensible context platform and MCP server designed to solve the critical problem of context overload for AI coding agents. It intelligently distills and delivers highly relevant code snippets, reducing token waste by 60-80% and preventing agents from getting overwhelmed by irrelevant information. By offering a single natural language interface for code discovery, CodeWeaver empowers agents to focus on task execution rather than tool selection, improving both search precision and development efficiency.

主な機能

• 4 GitHub stars
• Pinpoint accuracy with exact line/column references for code context.
• Agent-driven context curation via MCP sampling for enhanced precision without polluting main agent context.
• Background indexing with file watching ensures the code context remains current and relevant.
• Hybrid intelligence combining text, semantic, and AST analysis for superior search results.
• Extensible platform supporting over 10 embedding providers, multiple vector stores, and custom plugins.

ユースケース

• Rapidly understand unfamiliar codebases by providing precise, targeted context to AI agents.
• Significantly reduce AI token consumption and associated costs for coding agent interactions.
• Improve the accuracy and relevance of AI agent suggestions and prevent hallucinations.