Metadata-Version: 2.4
Name: OBRAG
Version: 0.1.2
Summary: Obsidian RAG assistant with access to your local vault and web search.
Author-email: Angel Patricio <aopatric@mit.edu>
License-Expression: MIT
Project-URL: Homepage, https://github.com/aopatric/obrag
Classifier: Development Status :: 4 - Beta
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: tomli-w<2.0,>=1.2.0
Requires-Dist: tqdm<5.0,>=4.67.1
Requires-Dist: langchain-chroma<0.3,>=0.2.4
Requires-Dist: langchain-openai<0.4,>=0.3.24
Requires-Dist: langchain-huggingface<0.4,>=0.3.0
Requires-Dist: langchain-community<0.4,>=0.3.26
Requires-Dist: langgraph<0.5,>=0.4.8
Requires-Dist: sentence-transformers<4.2,>=4.1.0
Requires-Dist: duckduckgo-search<9.0,>=8.0.4
Dynamic: license-file

# OBRAG
Obsidian-linked RAG Agent.

Installation:

Requires Python >= 3.11:

```python
pip install obrag
obrag --help
```

If there are any issues, reconfiguring with `obrag --reconfigure` should solve problems.

Simple Obsidian-linked RAG CLI for querying note base; API inference and local embeddings with HuggingFace.

To rebuild the vector store, use `obrag --rebuild`.

To skip the CLI and ask a question directly from the terminal, use `obrag --ask QUESTION`.

Happy querying.
