About ClawLab
Research-grade skills. Not a marketplace.
ClawLab is a curated skill store for OpenClaw agents. Every skill we ship is built by experts, grounded in peer-reviewed research, and validated with automated A/B testing.
OpenClaw has 247K+ GitHub stars and is overtaking every other agent framework. But the skill ecosystem is broken — ClawHub has 13,000+ free skills with no quality gate, and open marketplaces let anyone list anything.
We started ClawLab because we believe the highest-value skills require real expertise to build. Our first product — Metacognitive Calibration — synthesizes 135 peer-reviewed papers into a protocol that makes AI agents express confidence honestly. It took weeks of A/B testing to get right.
Every ClawLab skill follows the same standard: built on research, validated with automated testing, and designed for minimal token overhead. We release 2-3 new skills per week. Quality over quantity, always.
The team
Built by an AI systems engineer and a cognitive scientist
One side of ClawLab focuses on operational reliability. The other focuses on human and AI decision quality. The product is the overlap.
Shawkat Didar
Co-founder & AI Agent Systems
Builds the systems that make AI agents useful in practice — automated testing pipelines that catch behavioral regressions, skill architectures that stay under token budget, and delivery infrastructure that gets skills from lab to machine. Focus: making agentic AI work reliably outside of demos.
Ali Shahidy, Ph.D.
Co-founder & Behavioral Science
Cognitive scientist who designs the behavioral foundations behind ClawLab skills. Research background in how humans calibrate confidence and make decisions under uncertainty — the same principles that make an AI agent push back when it should instead of sounding certain about everything.