HomeProjectsArticlesPK BoardAbout
Explore Projects

LLocalSearch

Stale
GitHub Go Apache-2.0

Description

A completely locally running search aggregator using LLM agents. Users can ask questions and the system uses a chain of LLMs to find answers without any external API keys.

Key Features

  • Completely local operation with no API keys required — full privacy protection
  • Recursive LLM agent chain for multi-step internet search and reasoning
  • Live logs and source links providing transparency into agent reasoning
  • Supports follow-up questions for iterative research
  • Mobile-friendly responsive design with dark and light mode

Use Cases

💡 Privacy-respecting web search without relying on commercial search engines
💡 Research tasks requiring multi-source information aggregation
💡 Educational exploration of how LLM agents use tools recursively
💡 Local knowledge base queries using self-hosted language models

Strengths & Limitations

✅ Strengths

  • • High community interest (5.9k stars)
  • • Permissive open-source license (Apache-2.0)
  • • Established track record (2 years in production)
  • • Responsive to issues, low backlog

Tags

go agent rag llm tools

Categories

📚 RAG Tools

Quick Start

Clone the repo, run `docker-compose up -d`. Ensure Ollama is running with a supported model. Configure `.env` if Ollama runs on a different device. Access the web UI and start asking questions.
Visit GitHub

Related Projects

Prompt Engineering Guide

78.0k · MDX
Stale A

Comprehensive guides, papers, lessons, notebooks and resources for prompt engineering, context engineering, RAG, and AI Agents.

prompt-engineeringragagent +2
  • · Prompt engineering techniques - systematic coverage of Zero-Shot, Few-Shot, Chain-of-Thought, Tree of Thoughts and other core methods
  • · RAG and tool-use guides - detailed coverage of Retrieval Augmented Generation, ART, ReAct and other reasoning-with-tools techniques
  • · Model-specific guides - dedicated prompting strategies for ChatGPT, GPT-4, LLaMA, Gemini, Mistral, Code Llama and more
View full details → View on GitHub

WeKnora

21.2k · Go
Active A

Tencent's open-source LLM-powered framework for deep document understanding, semantic retrieval, and context-aware answers using RAG.

goragllm +3
  • · RAG-based Quick Q&A for everyday document lookups with semantic retrieval
  • · ReAct Agent autonomously orchestrating retrieval, MCP tools, and web search for complex multi-step tasks
  • · Wiki Mode where agents distill raw documents into self-maintaining, interlinked markdown knowledge bases with interactive knowledge graphs
View full details → View on GitHub

Scira

11.9k · TypeScript
Active A+

A minimalistic AI-powered search engine that helps you find information on the internet and cites it too. Powered by Vercel AI SDK.

typescriptllmrag +3
  • · Agentic research platform with multi-step planning that breaks complex questions into sub-tasks
  • · 17 search modes including Web, Academic, GitHub, Reddit, Stocks, Crypto, YouTube, and X (Twitter)
  • · 28 integrated tools for search, financial data, location, media, code execution, and file query
View full details → View on GitHub

MemAgent

1.1k · Python
Stale B

A MemAgent framework that can extrapolate to 3.5M context tokens, along with a training framework for RL training of any agent workflow.

memoryagentrag +2
  • · Ultra-Long Context Processing: Extrapolate from 8K training context to 3.5M tokens with performance loss under 5%
  • · Reinforcement Learning Driven: Trained with RLVR (Reinforcement Learning from Verifiable Rewards), extends DAPO algorithm for end-to-end multi-turn conversation optimization
  • · Linear Time Complexity: Breaks through computational bottlenecks in long-text processing with linear resource scaling
View on GitHub