The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script, scripts/check_applicant_fit.py, to match applicant details against job requirements. This script is local to the skill, uses only standard Python libraries, and follows a deterministic processing model that avoids external network calls or untrusted dependencies.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content provided by job applicants. To mitigate potential injection attacks, the skill provides specific instructions to the agent to distinguish between applicant claims and confirmed facts, use pseudonyms, and adhere to a strict reporting format that separates internal data from external communications.\n- [SAFE]: The skill exhibits high security maturity by incorporating programmable guardrails within its scripts to detect and block the accidental exposure of internal notes, other candidate information, or personal attributes unrelated to the job. It also includes comprehensive guidelines for PII management, emphasizing minimal data handling and secure storage practices.

applicant-reply