The Agent Skills Directory

[SAFE]: The skill provides instructional guidance for legal and technical analysis of penetration testing terms. No code execution or sensitive data access is involved.
[PROMPT_INJECTION]: No attempts to override safety guidelines, ignore previous instructions, or extract system prompts were found. The language is professional, instructional, and domain-specific.
[DATA_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or unauthorized network operations were identified. The term 'exfiltration' is used strictly within the context of defining the scope of a penetration test (e.g., asking which data is allowed to be exfiltrated by the tester).
[OBFUSCATION]: No encoded content, hidden characters, or homoglyph attacks were detected. The use of 'ae' and 'ue' in German words (e.g., 'traegt', 'bestimmungsgemaesse') is a standard alternative to umlauts and not a sign of obfuscation.
[REMOTE_CODE_EXECUTION]: The skill does not perform any remote code downloads or execution. References to legal databases like gesetze-im-internet.de are for manual verification of norms.
[DYNAMIC_EXECUTION]: No runtime code generation, script injection, or unsafe deserialization patterns are present.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests user input regarding a case or contract, it lacks any high-privilege capabilities (like file writing or network requests) that could be exploited by malicious data.

penetration-test-per-seat-platform-dsa