The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill's behavior is consistent with its stated purpose of auditing and normalizing repository content.
[PROMPT_INJECTION]: As a meta-skill that processes other skill files, it contains an inherent surface for indirect prompt injection. This is a characteristic of auditing tools rather than a vulnerability, but it remains a relevant surface for data ingestion.
Ingestion points: agent.yaml, README.md, and SKILL.md files within the provided skills_directory (SKILL.md).
Boundary markers: No explicit delimiter or instructions to ignore embedded commands are mentioned in the reading logic.
Capability inventory: File system read and write access for updating metadata and generating the ASQM_AUDIT.md report.
Sanitization: No specific filtering or validation mechanisms for the content being processed are described.

curate-skills