---
title: LLM Reviewer Foundry
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 4.44.0
python_version: '3.10'
app_file: app.py
pinned: false
emoji: 📚
short_description: LLM workspace for all your deduplication and screening needs
---

# Hugging Face Space Bundle (Azure Foundry)

This folder contains a self-contained Space app:
- `app.py` (Gradio entrypoint)
- `reference_parser.py` (reference parsing + dedup)
- `screen_excel_foundry_safe.py` (Azure Foundry screening)
- `requirements.txt`

## Required Space Secrets
Set these in your Hugging Face Space settings:
- `AZURE_FOUNDRY_ENDPOINT`
- `AZURE_FOUNDRY_DEPLOYMENT`
- `AZURE_API_KEY2`
- `APP_USERNAME` (login username for the app)
- `APP_PASSWORD` (login password for the app)

No Foundry endpoint/deployment/key values are hardcoded.

## Optional Safety Environment Variables
- `MAX_SCREENING_ROWS` (default `3000`)
- `MAX_TITLE_CHARS` (default `1200`)
- `MAX_ABSTRACT_CHARS` (default `8000`)
- `MAX_UPLOAD_FILES` (default `20`)
- `MAX_UPLOAD_FILE_MB` (default `25`)
- `MAX_CRITERIA_FILE_MB` (default `2`)
- `APP_TMP_DIR` (default `./artifacts`) for generated downloadable files
- `DOWNLOAD_LINK_TTL_SECONDS` (default `86400`) for signed download-link expiration

## Local Run
```bash
cd hf_space_foundry
pip install -r requirements.txt
python app.py
```

## Dependency note
`gradio==4.44.0` requires a pre-1.0 `huggingface_hub` API (it imports `HfFolder`).
This repo pins `huggingface_hub==0.25.2` in `requirements.txt` to avoid startup import errors.