Skip to content

SearchMuse

Intelligent web research powered by local LLMs.

SearchMuse is a CLI tool that performs iterative web research using LLM-powered search refinement and transparent source citation. It combines DuckDuckGo search, web scraping, content extraction, and LLM synthesis to produce well-cited answers to research queries.

Features

  • Iterative search refinement — LLM generates search strategies, assesses coverage, and refines until sufficient
  • Multi-provider LLM support — Ollama (local), Claude, OpenAI, Gemini
  • Transparent citations — Every claim is backed by numbered sources with URLs
  • Interactive REPL — Chat-like terminal interface with Rich formatting
  • Hexagonal architecture — Clean separation of ports and adapters
  • Fully typed — Strict MyPy compliance with PEP 561 marker

Quick Start

# Install
pip install searchmuse

# Start Ollama (if using local LLM)
ollama serve

# Pull a model
searchmuse ollama pull mistral

# Interactive mode
searchmuse

# Single query
searchmuse search "What are the latest developments in quantum computing?"

Documentation