---
title: ScholarFetch Web
emoji: 📚
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 4.44.1
python_version: "3.11"
app_file: app.py
pinned: false
---

# ScholarFetch Web (Hugging Face Space)

This Space installs `scholarfetch` and exposes a web UI for:
- Multi-engine search (keywords, author names, DOI)
- Author disambiguation and author-paper traversal
- Abstract and full-text reading for selected papers
- Reference expansion and one-click reference resolution
- Stateful saved-paper collections
- Export of saved collections as `citations`, `abstracts`, `bib`, or `fulltext`
- Public MCP streamable-http endpoint at `/mcp`

Runtime note:
- Python is pinned to `3.11` to avoid Python 3.13 `audioop` compatibility issues in transitive Gradio dependencies.
- `huggingface_hub` is pinned to `0.25.2` for compatibility with `gradio==4.44.1` (`HfFolder` import path).
- `fastapi` and `uvicorn` are pinned to versions compatible with `gradio==4.44.1` to avoid template-rendering crashes on Space startup.

## Required Space Secrets
Set these in **Settings -> Variables and secrets**:
- `ELSEVIER_API_KEY`
- `ELSEVIER_INSTTOKEN` (optional)
- `SPRINGER_META_API_KEY`
- `SPRINGER_OPENACCESS_API_KEY`

## Local run
```bash
pip install -r requirements.txt
python app.py
```

## MCP Endpoint (for Langflow / MCP clients)
Use the public Space URL with `/mcp`, for example:

`https://laibniz-scholarfetch-web.hf.space/mcp`

If your client needs Streamable HTTP, use this URL directly in the MCP server config.

## UI Workflow
Suggested flow:
1. Search for a topic or DOI
2. Read abstracts or full text
3. Expand references or author branches
4. Save promising papers into a named collection
5. Export the collection as a research packet