The Agent Skills Directory

[COMMAND_EXECUTION]: Executes a local shell script scripts/portfolio-scan.sh to collect metrics across the skill directory.
[COMMAND_EXECUTION]: Performs file system operations including global searches with grep and file removal or writing (DELETE/MERGE actions) when executing optimization recommendations.
[PROMPT_INJECTION]: Ingests and processes content from external SKILL.md files and CLAUDE.md which could contain adversarial instructions intended to influence the auditor's recommendations.
Ingestion points: Reads all SKILL.md files in the portfolio, the global CLAUDE.md, and settings.json during the audit and dependency check phases.
Boundary markers: The workflow does not specify the use of delimiters or 'ignore instructions' warnings when analyzing the content of other skills.
Capability inventory: Includes shell script execution, cross-file searching, and the ability to delete or modify files in the local skills directory.
Sanitization: There is no mention of sanitizing or escaping the text contents of the skills being reviewed before they are processed by the agent.

skill-portfolio-review