Skip to content

Dataset · OpenClaw

Security and trust signals for AI agent skills

OpenClaw/clawhub-security-signals

Tabular-text security assessment dataset for public OpenClaw skills with scanner verdicts.

What's inside

A cleaned MIT-licensed dataset released with a paper on assessing the security of public skills in the OpenClaw agent registry. Includes 10 to 100 thousand English records in tabular and text format. The data reflects trust checks, code provenance, and analysis results via VirusTotal, static analyzers, and SkillSpector.

What you can do with it

  • Training classifiers to automatically detect malicious plugins and AI agent skills
  • Analyzing discrepancies between static code analysis and antivirus scanners
Tasks
Text classification
Languages
English
Size
10K–100K rows
Downloads this month
693
Likes
53

How to load it

from datasets import load_dataset

ds = load_dataset("OpenClaw/clawhub-security-signals")