The Agent Skills Directory

[SAFE]: The skill is a robustly designed tool for financial risk analysis, following a structured methodology that includes knowledge source tracing and quality validation. It enforces strict replication of source text to prevent AI hallucinations and explicitly prohibits subjective risk grading to maintain objectivity.
[DATA_EXFILTRATION]: The skill contains comprehensive data handling rules, including mandatory desensitization for sensitive information such as bank accounts, IDs, and customer names. No unauthorized external network calls or data exfiltration patterns were identified.
[PROMPT_INJECTION]: Instructions include strict output constraints and a dedicated self-check step (Step 5) to verify the integrity and quantity of extracted risk points, which serves as an effective control against manipulation through malicious content within ingested documents.
[REMOTE_CODE_EXECUTION]: No remote code execution or suspicious dependency loading was found. The provided Python validation script is a local structural checker that uses standard libraries and safe logic to verify report formatting.

credit-risk-extraction