Back to skill
Skillv1.0.0

ClawScan security

Bash 1.0.2 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 4, 2026, 11:30 AM
Verdict
benign
Confidence
high
Model
gpt-5-mini
Summary
This is a documentation-only skill that provides Bash best-practices and examples; its declared requirements and files are coherent with that purpose.
Guidance
This skill is documentation-only and appears to be what it claims: Bash best-practices and examples. Because it is instruction-only, installing it won’t download code or request credentials. Before installing, verify you trust the publisher (there is a small ownerId mismatch in the packaged _meta.json vs. registry metadata), and be cautious about copying example commands into a shell without review—examples may operate on files or have side effects. If you want extra caution, review the .md files yourself or test any example commands in a disposable/sandbox environment.

Review Dimensions

Purpose & Capability
noteName/description match the provided markdown guidance and the only required binary (bash) is appropriate. Minor metadata inconsistency: the registry metadata Owner ID (kn7572...) does not match the ownerId in _meta.json (kn73vp5...), which is likely a packaging/metadata error but not evidence of malicious behavior.
Instruction Scope
okSKILL.md and the included .md files contain static guidance and example commands for writing Bash safely. They do not instruct the agent to read arbitrary system files, exfiltrate data, or fetch external endpoints. Examples show common shell commands but are presented as examples, not runtime directives to gather user data.
Install Mechanism
okNo install spec and no code files—this is instruction-only, which minimizes risk because nothing will be downloaded or written to disk.
Credentials
okThe skill requires no environment variables, secrets, or config paths. The single required binary (bash) is appropriate for its purpose.
Persistence & Privilege
okalways is false and the skill is user-invocable. It does not request persistent system presence or special privileges and does not modify other skills or system-wide agent settings.