Metadata-Version: 2.3
Name: litcurator
Version: 0.0.3
Summary: litcurator: LLM-powered pubmed literature extraction
Author: Eric Thomson
Author-email: Eric Thomson <thomson.eric@gmail.com>
License: Apache-2.0
Requires-Dist: requests
Requires-Dist: anthropic
Requires-Dist: python-dotenv
Requires-Dist: streamlit
Requires-Python: >=3.12
Project-URL: Repository, https://github.com/EricThomson/litcurator
Description-Content-Type: text/markdown


# litcurator 

An LLM-based literature filter. 

Litcurator works in two stages. First, retrieve publications from PubMed within your field of interest. This casts a very broad net (e.g., neuroscience). Then, narrow them down to a final curated list using an LLM that builds up a profile of your interests. It uses [streamlit](https://streamlit.io/) to display articles in the browser.

Initial focus is on systems neuroscience. 

## API Keys Needed
For this to work you need some API keys that you should store in `.env`:
- An NCBI (National Center for Biotechnology Information) API key. For info on this: https://www.ncbi.nlm.nih.gov/datasets/docs/v2/api/api-keys/. 
- An API key for an LLM vendor (I'm currently using anthropic). 

## Status
- Tools for labeling and displaying articles are done.
- Using them to generate ground truth labels for building user profile. 

