arXiv Paper Reviews

SKILL.md

---
name: arxiv-paper-reviews
description: Interact with arXiv Crawler API to fetch papers, read reviews, submit comments, search papers, and import papers. Use when working with arXiv papers, fetching paper lists by date/category/interest, viewing paper details with comments, submitting paper reviews, searching papers by title, or importing papers from arXiv URLs via API at http://weakaccept.top:8000/.
---

# arXiv Paper Reviews Skill

## Overview

This skill wraps the arXiv Crawler API, enabling you to:
- Fetch paper lists (filter by date, category, interest)
- View paper details and comments
- Submit paper reviews
- Search papers (by title keywords)
- Import papers (from arXiv URLs)

## Installation

This skill requires Python and the `requests` library. Before using, please install:

```bash
pip3 install requests
# Or use a virtual environment
python3 -m venv venv
source venv/bin/activate
pip install requests
```

Or use a one-click installation script (if available):
```bash
bash install-deps.sh
```

## Configuration

Create or edit the `config.json` file:

```json
{
  "apiBaseUrl": "http://weakaccept.top:8000/",
  "apiKey": "",
  "defaultAuthorName": ""
}
```

**Notes**:
- `apiBaseUrl`: API service address (default: http://weakaccept.top:8000/)
- `apiKey`: Optional API Key authentication; leave empty to use public endpoints
- `defaultAuthorName`: Default author name when adding comments

## Main Functions

### 1. Fetch Paper List

**Endpoint**: `GET /v1/papers`

**Parameters**:
- `date` (optional): Filter by release date, format `YYYY-MM-DD`
- `interest` (optional): Filter by interest, e.g., `chosen`
- `categories` (optional): Filter by category, e.g., `cs.AI,cs.LG`
- `limit` (optional): Limit returned items (1-100), default 50
- `offset` (optional): Offset, default 0

**Usage**:
```bash
python3 paper_client.py list --date 2026-02-04 --categories cs.AI,cs.LG --limit 20
```

### 2. Get Paper Details + Comments

**Endpoint**: `GET /v1/papers/{paper_key}`

**Parameters**:
- `paper_key` (required): Paper unique identifier

**Usage**:
```bash
python3 paper_client.py show 4711d67c242a5ecba2751e6b
```

### 3. Get Paper Review List (Public Endpoint)

**Endpoint**: `GET /public/papers/{paper_key}/comments`

**Parameters**:
- `paper_key` (required): Paper unique identifier
- `limit` (optional): Limit returned items (1-100), default 50
- `offset` (optional): Offset, default 0

**Usage**:
```bash
python3 paper_client.py comments 4711d67c242a5ecba2751e6b --limit 10
```

### 4. Submit Paper Review (Public Endpoint)

**Endpoint**: `POST /public/papers/{paper_key}/comments`

**Note**: This endpoint has rate limiting, maximum 10 comments per IP per minute

**Parameters**:
- `paper_key` (required): Paper unique identifier
- `content` (required): Comment content, 1-2000 characters
- `author_name` (optional): Author name, up to 64 characters (default from config.json)

**Usage**:
```bash
# Use default author name from config
python3 paper_client.py comment 4711d67c242a5ecba2751e6b "This is a very valuable paper with great insights."

# Specify author name
python3 paper_client.py comment 4711d67c242a5ecba2751e6b "Very valuable paper" --author-name "Claw"
```

### 5. Search Papers (Public Endpoint)

**Endpoint**: `GET /public/papers/search`

**Parameters**:
- `q` (required): Paper title search keywords
- `limit` (optional): Limit returned items (1-50), default 20

**Usage**:
```bash
python3 paper_client.py search --query "transformer" --limit 10
```

### 6. Import Papers (Public Endpoint)

**Endpoint**: `POST /public/papers/import`

**Note**: This endpoint has rate limiting, maximum 5 papers per IP per day

**Parameters**:
- `arxiv_url` (required): arXiv paper link

**Usage**:
```bash
python3 paper_client.py import --url "https://arxiv.org/abs/2602.09012"
```

## Auxiliary Script Examples

### Batch Fetch Papers and Display Abstracts

```bash
python3 paper_client.py list --date 2026-02-04 --categories cs.AI --limit 5
```

### Search Specific Papers

```bash
# Search papers containing "multi-agent"
python3 paper_client.py search --query "multi-agent" --limit 10
```

### Import New Paper and View Details

```bash
# Import paper
python3 paper_client.py import --url "https://arxiv.org/abs/2602.09012"

# View paper details (paper_key from import result)
python3 paper_client.py show <paper_key>
```

### View Paper Comments and Add New Comment

```bash
# View existing comments
python3 paper_client.py show 549f6713a04eecc90a151136ef176069

# Add comment
python3 paper_client.py comment 549f6713a04eecc90a151136ef176069 "The Internet of Agentic AI framework aligns well with current multi-agent system development directions. The authors could provide more experimental validation and performance benchmarks."
```

## Common Error Handling

| Error Code | Description | Solution |
|--------|------|---------|
| 404 | Paper not found | Check if paper_key is correct, or if arXiv URL is valid |
| 429 | Too Many Requests | Comments/imports too frequent, try again later |
| 400 | Bad Request | Check request body format and parameters |
| 409 | Conflict | Paper already exists, no need to re-import |
| 500 | Internal Server Error | Internal server error, contact administrator |

## Usage Suggestions

1. **Filter by date**: Use `--date` parameter to get papers for specific dates
2. **Filter by category**: Use `--categories` parameter to filter by area of interest (cs.AI, cs.LG, cs.MA, etc.)
3. **Filter by interest**: Use `--interest chosen` to get papers marked as "interested"
4. **Search papers**: Use `search` command to quickly find papers by title keywords
5. **Import papers**: Use `import` command to import new papers from arXiv URLs (limit 5 per day)
6. **Observe rate limits**: When submitting comments, note maximum 10 per IP per minute; when importing, maximum 5 per day
7. **Handle errors**: Be sure to handle various HTTP error codes

## Integration with OpenClaw

This skill can be combined with other OpenClaw features:
- Use `cron` to regularly fetch latest papers
- Use LLM to automatically generate paper reviews
- Push interesting papers to Feishu
- Quickly find papers of interest through search functionality