skill-security-analyzer

You can identify security risks in Claude skills by validating frontmatter metadata, analyzing SKILL.md instructions for prompt-injection risks, and inspecting bundled scripts and assets against security patterns to generate an executive summary and actionable checklist.

A prompt-injection risk in Claude skills is a safety vulnerability hidden within instructions. It is detected through systematic SKILL.md instruction analysis that reviews content clarity and flags potential prompt-injection threats before safe deployment.

Validating SKILL.md frontmatter metadata involves checking for the required presence of the skill name and description. This preflight security analysis ensures required metadata is present before proceeding to instruction and bundle reviews.

A security bundle inspection for Claude skills covers a comprehensive review of bundled scripts, references, and assets. It surfaces security concerns by inspecting these bundled resources against established security patterns.

Yes, this security analyzer works for skills without bundled resources by focusing the assessment on frontmatter validation and SKILL.md instruction analysis to identify safety issues and prompt-injection risks.

The best way to generate a security checklist for Claude skill deployment is running a structured security analysis that outputs executive summaries and findings with severity levels based on frontmatter, instruction, and bundle inspections.