skills-audit

Audit local agent skills for security by parsing each skill's SKILL.md frontmatter and inspecting bundled resources to produce a verdict and numeric risk score. This surfaces compliance risks and toxic content before you deploy.

A skill security audit checks frontmatter for required name and description declarations while inspecting bundled scripts and assets. It reports dependencies, components, and potential toxic content detected during the resource analysis.

Yes, you can scan a single skill folder instead of an entire skills directory. The audit applies to a configured skills root or an individual skill folder, producing a safe, suspicious, or unsafe verdict for either target.

The skill risk score is calculated during a compliance audit by evaluating frontmatter validation and resource inspection results. It generates a numeric score alongside a verdict, highlighting concrete evidence for each security finding.

The best way to identify toxic content in bundled agent skills is to run a resource-aware audit that inspects bundled scripts, references, and assets. This process detects potential toxic content and reports it with remediation guidance.

Your skill audit reports a missing frontmatter validation error because every valid skill must declare a name and description in its SKILL.md frontmatter. The audit identifies these missing declarations and flags them as compliance violations.