Agent Fusion
Empower AI agents with instant, hallucination-free access to your codebase and documentation through an intelligent local semantic search engine.
概要
Agent Fusion provides a robust local semantic search engine designed to give AI agents immediate access to your entire codebase, documentation, and various file types, eliminating hallucinations and ensuring accurate responses. It features a powerful Context Engine that indexes code, Markdown, Word, and PDFs, enabling AI assistants to understand concepts rather than just keywords through hybrid search. Optionally, its Task Manager component facilitates multi-agent task orchestration, allowing AIs to collaborate, vote on decisions, and track progress, making it an ideal solution for grounding AI development workflows and fostering collaborative intelligence.
主な機能
- Optional multi-agent Task Manager for task orchestration, collaborative decision-making, and audit trails.
- 31 GitHub stars
- Runs 100% locally with a lightweight embedding model included, easily deployable via a single JAR.
- Local semantic search engine for code, documentation (Markdown, Word, PDF), and other files.
- Hybrid search combining semantic, full-text, symbol, and Git history for comprehensive results.
- Allows AI agents to query codebase context directly via natural language prompts like 'use query_context'.
ユースケース
- Enabling AI coding assistants to accurately query project code and documentation without hallucinations.
- Orchestrating collaboration and consensus among multiple AI agents for complex development tasks.
- Providing developers with conceptual search capabilities across their codebase to understand logic beyond keywords.
