huri-module-rag

Retrieval-augmented answers with long-term memory (Qdrant + an OpenAI-compatible LLM), and the question aggregator.

pip install -e 'packages/huri-module-rag[rag]'   # from a huri-modules checkout, next to huri-core and huri-events
HURI_MODULES=rag,qag ...   # enable it on the server

rag and the ingestion and memory tools (python -m huri_rag.ingestion, memory_inspect, memory_maintenance) need the rag extra (Qdrant client, sentence-transformers, httpx, pypdf, semantic_chunker); qag only joins events and needs none of it, so a host that runs qag alone installs the package without the extra ([tool.huri-modules.extras] in pyproject.toml; huri hub install and scripts/install_local.sh add it). Hardware pins of rag: requirements-amd.rag.txt (install it before the package).

A distribution of huri-modules, versioned in its pyproject.toml (see RELEASING.md there). It depends on huri-core and huri-events, which are not on PyPI: without a checkout, install them from git in the same pip command, so pip never looks them up on an index:

ref=0.1.0   # a huri-modules release tag
core=0.4.0  # HURI_CORE_REF in that release's versions.env
pip install \
  "huri-core[server] @ git+https://github.com/Lucy-Robotics/huri-core.git@$core" \
  "huri-events @ git+https://github.com/Lucy-Robotics/huri-modules.git@$ref#subdirectory=packages/huri-events" \
  "huri-module-rag[rag] @ git+https://github.com/Lucy-Robotics/huri-modules.git@$ref#subdirectory=packages/huri-module-rag"