The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection due to how it handles external, untrusted content. ● Ingestion points: The analyze, decode, and resume commands ingest text from interview transcripts, job descriptions, and external resumes. ● Boundary markers: The skill does not define clear boundaries or 'ignore' instructions for the data it processes. ● Capability inventory: The agent is granted WebSearch, WebFetch, and file system Read/Write capabilities (specifically for the coaching_state.md file). ● Sanitization: There is no evidence of logic to sanitize or escape inputs before they are processed by the language model.
[EXTERNAL_DOWNLOADS]: The skill implements a research command that uses WebSearch and WebFetch to gather company intelligence from various public websites, blogs, and news sources to provide context for the user's interview preparation.

interview-coach