---
name: semantic-scholar
description: Search Semantic Scholar's 200M+ paper corpus, traverse citation graphs, get paper recommendations, and export BibTeX. Use when finding papers that cite a given work, exploring what a paper references, getting AI-powered paper recommendations, or searching academic literature with citation count and year filters. Complements alphaXiv (paper Q&A) and PubMed (biomedical focus) with broader coverage and citation network analysis.
---

# Semantic Scholar via s2cli

## Overview

Semantic Scholar aggregates 200M+ papers from PubMed, arXiv, ACL, DBLP, and more. Use `s2cli` for citation graph exploration, paper recommendations, and filtered academic search. It auto-outputs JSON when piped, making it agent-friendly.

## Installation

```bash
pip install s2cli
```

No API key required for basic use. For higher rate limits, set:
```bash
export S2_API_KEY=your_key  # Get from https://www.semanticscholar.org/product/api
```

## Commands

### Search papers

```bash
# Basic search
s2cli search "transformer scaling laws"

# With filters
s2cli search "CRISPR gene editing" --year 2022- --min-citations 50

# Open access only
s2cli search "machine learning drug discovery" --open-access

# JSON output (for agent parsing)
s2cli search "attention mechanisms" --json
```

### Get paper details

```bash
# By arXiv ID
s2cli paper ARXIV:1706.03762

# By DOI
s2cli paper DOI:10.1038/s41586-021-03819-2

# By Semantic Scholar Corpus ID
s2cli paper CorpusId:215416146

# Full JSON
s2cli paper ARXIV:1706.03762 --json
```

### Citation graph traversal

```bash
# What papers cite this work?
s2cli citations ARXIV:1706.03762

# What does this paper reference?
s2cli references ARXIV:1706.03762

# Filter citations by year
s2cli citations ARXIV:1706.03762 --year 2023-
```

This is the key differentiator — trace the influence of a paper forward (who cited it) and backward (what it built on).

### Paper recommendations

```bash
# Get recommendations based on a paper
s2cli recommendations ARXIV:1706.03762
```

Uses Semantic Scholar's recommendation engine to suggest related papers.

### Export BibTeX

```bash
s2cli bibtex ARXIV:1706.03762
```

### Author search

```bash
# Find an author
s2cli author search "Geoffrey Hinton"

# List author's papers
s2cli author papers <author_id>
```

## Paper ID formats

s2cli accepts multiple ID formats:
- `ARXIV:2301.00001` — arXiv ID
- `DOI:10.1038/...` — DOI
- `CorpusId:12345` — Semantic Scholar corpus ID
- `PMID:12345678` — PubMed ID
- `ACL:P18-1234` — ACL Anthology ID

## When to use Semantic Scholar vs other tools

| Need | Use |
|---|---|
| Read a paper's full text, ask questions about it | `alpha` (alphaXiv) |
| Search biomedical literature specifically | `pubmed-database` skill |
| Search 240M+ papers with bibliometric analysis | `openalex-database` skill |
| Traverse citation graphs, get recommendations | `s2cli` (this skill) |
| Current events, products, non-academic | `web_search` |

## Workflow integration

For literature reviews, combine tools:
1. `s2cli search` to find seed papers
2. `s2cli citations` + `s2cli references` to map the citation network
3. `alpha get` to read key papers in detail
4. `alpha ask` to query specific claims
5. Synthesize findings into the research artifact