The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill generates hiring artifacts by processing user responses to an interactive 10-question interview, creating a surface for potential instruction injection by malicious users.
Ingestion points: User answers provided during the interview workflow defined in SKILL.md.
Boundary markers: Absent; there are no instructions to use delimiters or to treat user responses as data rather than instructions.
Capability inventory: The skill produces text-based artifacts (scorecards, loops, sourcing plans). It does not contain capabilities for arbitrary command execution, network exfiltration of sensitive files, or file system modifications.
Sanitization: Absent; the skill does not perform validation or filtering on user-provided responses before interpolation into final documents.
[PERSISTENCE]: The skill includes instructions to store generated artifacts in 'persistent memory' for reuse in future hiring cycles. This is a legitimate platform feature for state management and does not constitute a malicious persistence attack vector (e.g., shell profiles or registry modifications).
[EXTERNAL_DOWNLOADS]: The skill references accessing industry-specific resources like the 'Partnership Leaders' compensation database and job boards for live data. These are established, well-known services for the partnerships field, and their use is safe and relevant to the skill's primary functionality.

partnerships-hiring