Get Hunter-RAG

Install

pip install hunter-rag
export OPENAI_API_KEY=...        # or any OpenAI-compatible endpoint:
export OPENAI_BASE_URL=http://localhost:11434/v1

Python 3.10 or later. Bring your own model: OpenAI, or a local or gateway endpoint that speaks the same API.

Ask a document

cdu paper.pdf                    # parse it, then ask
> Which finding had the highest specificity?
> /cite                          # the units behind the answer
> /trace                         # every step the agent took

PDF, Markdown, HTML, text, LaTeX, XML.

A whole corpus

cdu index ./docs                 # a folder or an Obsidian vault
cdu sql "SELECT element_type, count(*) FROM cdus GROUP BY 1"

From your coding agent

Any agent that can run a shell command can use it. The one-shot form prints the answer and exits:

cdu report.pdf -p "List the material risk factors, with citations."

A skill for Claude Code and similar agents ships in the repository (skills/hunterrag-corpus): it teaches the agent to cite unit addresses, not line numbers.

For your organisation

Hunter-RAG is built for enterprise and industrial use. Arc's engineers deploy it into your environment, with your documents and your models, and support it in production. Talk to Arc.

Personal, research and education use is free, as is a 90-day evaluation. Terms: Licensing.