Six PaperClaws Walk Into a Library...
If you search "PaperClaw" on GitHub, you'll find six different projects — all independently built, all focused on academic paper workflows, all named PaperClaw. Welcome to the Claw ecosystem's most crowded name.
But here's the thing: they're not interchangeable. Each one solves a genuinely different problem. This guide helps you pick the right one in under 2 minutes.
Quick Decision Tree
What do you need?
- "Generate an expert agent for my specific research field" → guhaohao0991
- "A full skill library for my research team" → meowscles69
- "Daily paper digests in my inbox" → PigeonDan1
- "Production skill library with team workflows" → 0xMerl99
- "Automated arXiv pipeline with relevance scoring" → AkaliKong
- "Three-layer autonomous research system" → 1692775560
The Comparison
| guhaohao0991 | meowscles69 | 0xMerl99 | PigeonDan1 | AkaliKong | 1692775560 | |
|---|---|---|---|---|---|---|
| Stars | 173 | 132 | 51 | 28 | 21 | 16 |
| Language | Python | Markdown | Markdown | Python | Python | Python |
| Type | Framework | Skill library | Skill library | Digest tool | Pipeline | Agent system |
| Best for | Generating domain experts | Research teams | Team workflows | Daily paper emails | arXiv tracking | Autonomous research |
1. PaperClaw (guhaohao0991) — The Domain Expert Generator
GitHub: guhaohao0991/PaperClaw · 173 stars
This is the most unique PaperClaw. Instead of being a tool itself, it's a framework that generates specialized paper-expert agents for any research domain you specify.
How it works:
- You specify a research field (e.g., "spatial transcriptomics" or "protein folding")
- It runs an 8-step workflow to generate a custom
AGENT.md - The generated agent knows your field's keywords, scoring dimensions, and evaluation criteria
- You get a permanent domain expert that can search, review, and critique papers in your niche
The demo uses Scientific ML and 3D geometry surrogate modeling as examples, but the generator works for any field.
Install: clawhub install paper-expert-generator
Best for: Researchers who want a permanent AI expert tuned to their specific subfield. If you're deep in one domain and read papers in it every week, this is your pick.
2. PaperClaw (meowscles69) — The Team Skill Library
GitHub: meowscles69/PaperClaw · 132 stars
27 production-ready skills organized into 5 categories:
| Category | Skills | Examples |
|---|---|---|
| Literature (6) | Systematic search, review synthesis, gap identification | |
| Synthesis (5) | Cross-paper analysis, hypothesis versioning | |
| Collaboration (5) | Lab knowledge handoffs, onboarding guides | |
| Output (6) | Manuscript drafting, grant writing, figure planning | |
| Tracking (5) | Citation monitoring, conference deadline tracking |
What makes it special: This is designed for research teams, not solo researchers. It includes skills for lab handoffs (when a postdoc leaves), shared hypothesis tracking, and collaborative grant writing. It's the "knowledge management layer" complementing LabClaw's biomedical execution toolkit.
Install: Copy individual skill folders to ~/.openclaw/skills/ or install all 27 at once.
Best for: Research labs and multi-person teams who need shared AI-assisted workflows.
3. PaperClaw (0xMerl99) — The Production Skill Set
GitHub: 0xMerl99/PaperClaw · 51 stars
Another 27-skill library, but focused on production-readiness and individual researcher workflows rather than team collaboration. Literature management, synthesis, manuscript output, and tracking — similar categories to meowscles69 but with different skill implementations.
Best for: Individual researchers who want a battle-tested skill collection without the team collaboration features.
4. paper_claw (PigeonDan1) — The Daily Digest Machine
GitHub: PigeonDan1/paper_claw · 28 stars
Completely different from the others. This is a standalone tool (not a skill library) that:
- Fetches papers from arXiv and other sources daily
- Classifies them by your configured research categories
- Generates multilingual AI summaries (supports DeepSeek, Kimi, OpenAI)
- Sends personalized digests straight to your inbox
- Runs on GitHub Actions — zero infrastructure needed
Two modes:
- For humans: Set up daily email digests for your research field
- For AI agents: Use as an OpenClaw skill for paper discovery
Install: Fork the repo, configure .env, enable GitHub Actions. Done.
Best for: Anyone who wants "morning paper briefing" in their email without lifting a finger. The most practical PaperClaw for daily use.
5. PaperClaw (AkaliKong) — The arXiv Pipeline
GitHub: AkaliKong/PaperClaw · 21 stars
An agent-orchestrated pipeline for the full academic research lifecycle:
- arXiv search with relevance scoring
- Deep reading and code evaluation
- Idea synthesis from multiple papers
Best for: Researchers focused on arXiv-heavy fields (CS, ML, physics) who want automated relevance filtering.
6. PaperClaw (1692775560) — The Three-Layer Brain
GitHub: 1692775560/PaperClaw · 16 stars
The most architecturally ambitious PaperClaw. Uses a three-layer design:
- Brain — LLM decision engine
- Hands — Skill execution layer
- Memory — Knowledge management system
Best for: Those interested in autonomous research agent architecture. Early stage but conceptually interesting.
Our Recommendation
| If you need... | Choose | Why |
|---|---|---|
| A permanent expert for YOUR field | guhaohao0991 | Generates a custom domain agent, not a generic tool |
| Team collaboration + lab handoffs | meowscles69 | Only PaperClaw designed for multi-person teams |
| Daily paper emails, zero setup | PigeonDan1 | Runs on GitHub Actions, delivers to your inbox |
| The most stars / community validation | guhaohao0991 (173 stars) | Most popular, actively maintained |
| arXiv-specific pipeline | AkaliKong | Focused on arXiv relevance scoring |
Editor's Pick: guhaohao0991 — The domain expert generator concept is genuinely novel. Most paper tools treat all fields the same; this one creates a specialist agent that understands your field's terminology, scoring criteria, and evaluation standards. That's a meaningful difference.
See the full side-by-side comparison: /compare/paperclaw
Last updated: March 23, 2026. All six projects are actively maintained.