Performs semantic code search for AI agents and CLI users, finding code by meaning rather than exact text matches.
Sponsored
Akin is a versatile semantic code search tool designed for both command-line users and AI agents via the Model Context Protocol (MCP). It operates entirely locally, eliminating the need for API keys or external services, utilizing local ONNX embeddings with the Nomic Embed Text v1.5 model. The tool creates a searchable index in a `.akin/` folder within your repository, intelligently chunking various file types like C#, Python, Markdown, and JavaScript for structure-aware indexing. It supports incremental updates in MCP server mode, ensuring the index remains current with file changes, making it an efficient solution for navigating large codebases semantically.
主な機能
01Semantic code search by meaning, not text
02Runs entirely locally with ONNX embeddings
03CLI tool and MCP server for AI agent integration
04Supports structure-aware chunking for multiple languages
05Automatic incremental indexing and re-indexing
060 GitHub stars
ユースケース
01Quickly find relevant code snippets in a repository using natural language queries
02Integrate with AI agents (like Claude Code) for context-aware code suggestions
03Automatically update code indexes for efficient, always-current semantic search
