Key Takeaways
- LabClaw provides 240 pre-built skills specifically for wet lab research workflows
- Covers PCR protocol design, CRISPR guide RNA selection, cloning strategy, cell culture protocols, and more
- Built on OpenClaw — inherits multi-platform messaging, MCP support, and skill marketplace access
- Skills encode domain-expert knowledge: proper controls, common pitfalls, troubleshooting steps
- Developed by wu-yc, open-source under MIT license
What Is LabClaw?
LabClaw is a wet-lab-focused AI agent built on OpenClaw. While most Claw ecosystem tools focus on computational biology, LabClaw bridges the gap to physical lab work — helping researchers design experiments, select protocols, and troubleshoot procedures.
Core attributes:
- Category: Wet lab research AI agent
- Skills: 240 pre-built lab-specific skills
- Base framework: OpenClaw (TypeScript)
- Repository: wu-yc/LabClaw
- License: MIT
Skill Categories
| Category | Skills | Examples |
|---|---|---|
| PCR & Primers | 28 | Protocol design, primer optimization, troubleshooting |
| Cloning | 22 | Restriction cloning, Gibson assembly, Golden Gate |
| CRISPR | 19 | Guide RNA design, off-target analysis, editing strategy |
| Cell Culture | 35 | Media selection, passaging protocols, contamination handling |
| Protein | 31 | Expression optimization, purification, Western blot |
| Sequencing | 18 | Library prep, quality assessment, data interpretation |
| Microscopy | 15 | Sample preparation, imaging parameters, analysis |
| General Lab | 72 | Safety, equipment, calculations, documentation |
What Makes Lab Skills Different
Lab skills differ from computational skills in a critical way: they encode physical-world knowledge that AI models often get wrong.
Example — PCR troubleshooting skill:
- Checks template concentration, primer Tm mismatch, extension time
- Suggests gradient PCR before adjusting all parameters
- Warns about common mistakes (expired reagents, wrong polymerase for GC-rich templates)
- Recommends proper controls (positive, negative, no-template)
This is knowledge that exists in experienced researchers' heads but rarely in textbooks.
FAQ
Q1: Do I need a coding background to use LabClaw?
No. LabClaw skills are triggered by natural language. Ask "help me design a PCR protocol for this gene" and the skill guides you through template, primers, conditions, and controls.
Q2: Can I use LabClaw skills with other agents?
Yes. Skills are standard SKILL.md files compatible with Claude Code, Codex, and any SKILL.md-compatible agent.
Q3: How do LabClaw skills differ from asking ChatGPT about lab protocols?
LabClaw skills encode expert knowledge about common pitfalls, proper controls, and troubleshooting steps. Generic LLMs may give plausible-sounding but incorrect protocol details.
Summary
LabClaw is the most comprehensive wet-lab skill set in the OpenClaw ecosystem. With 240 skills covering PCR, cloning, CRISPR, cell culture, protein work, and more, it bridges AI assistance from computational biology to physical lab work. The skills encode physical-world knowledge that AI models often get wrong — making it a practical tool for researchers at the bench.