Back to skill
Skillv1.1.0
ClawScan security
Robotics VLA · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMay 1, 2026, 8:31 AM
- Verdict
- Benign
- Confidence
- high
- Model
- gpt-5.5
- Summary
- This is a documentation-only robotics guidance skill with no code, install steps, credentials, persistence, or data-access behavior shown.
- Guidance
- This appears safe to install from an agentic-security perspective because it is only reference content. If applying the robotics advice to real robots, use normal engineering safety review and human supervision, but the skill artifacts themselves do not request dangerous runtime capabilities.
Review Dimensions
- Purpose & Capability
- okThe skill’s stated purpose—expert guidance on Vision-Language-Action robot foundation models—matches the included reference material on architecture, training, embodiments, and related work.
- Instruction Scope
- okThe instructions are informational and advisory; they do not direct the agent to run commands, access files, call APIs, modify systems, or take autonomous actions.
- Install Mechanism
- okNo install spec, package dependencies, required binaries, or executable code are present.
- Credentials
- okThe metadata declares no required environment variables, credentials, config paths, or OS-specific authority, which is proportionate for an instruction-only research guidance skill.
- Persistence & Privilege
- okThe artifacts show no persistence, background workers, credential/session use, memory storage, or privilege escalation mechanisms.