The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to process and review code authored by the user ("learner's own implementation"). This behavior introduces a surface for indirect prompt injection, as malicious instructions could be embedded within the reviewed code to influence the agent's behavior or the content of the generated reports.
Ingestion points: The skill reads project directories, git branches, and specific code artifacts during the review process as described in SKILL.md.
Boundary markers: The skill lacks explicit instructions for the agent to use delimiters or "ignore embedded instructions" warnings when processing the untrusted code being reviewed.
Capability inventory: The skill writes to .mentor/log.md, creates HTML report files, and generates PR comments.
Sanitization: No sanitization, escaping, or validation of the ingested code content is specified before it is incorporated into the agent's reasoning or the final reports.
[COMMAND_EXECUTION]: The instructions require the agent to invoke external skills such as /grilling and /ask-lonlydwolf for intake interviews and referrals. It also references a setup command /setup-faculty-skills. These are identified as resources associated with the author 'lonlydwolf'.
[COMMAND_EXECUTION]: The skill generates HTML reports (mentor-review-*.html) by reading and inlining content from local CSS and JavaScript files (.assets/course.js, .assets/palette.css). This involves dynamic generation and embedding of scripts within a browser-executable format based on local templates.

mentor