ClawHub Security Signals: 67,453 Public Agent Skill Versions

OpenClaw provides 67,453 sanitized, public agent-skill versions for security analysis. The dataset captures how a skill registry evaluates trust, provenance, bundled code, and scanner evidence at scale. It was last updated on June 1, 2026.

Use Cases

• Analyze trust signals across agent skills based on the described registry evaluation.
• Study provenance and code bundling patterns in public AI skills.
• Benchmark security scanner evidence for automated skill vetting.
• Train models to classify skill safety based on described security features.

Strengths

• Contains 67,453 latest public skill versions.
• Data is sanitized and released under an MIT license.
• Focuses on concrete security signals like trust and provenance evaluation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known but other scale metrics like file size are unknown.
• Freshness should be verified as the last update timestamp is from the future (2026-06-01).

Provenance

Source

OpenClaw

Collection Method

Captures evaluations from a public agent-skill registry.

Freshness

Last updated 2026-06-01 04:55:18