The Agent Skills Directory

[SAFE]: The skill consists of instructions and guidelines for a legal AI agent specializing in NIS2 (Network and Information Security Directive). It does not contain executable code, scripts, or commands that interact with the host system beyond basic text processing.
[DATA_EXFILTRATION]: No sensitive data access or network exfiltration patterns were detected. The skill references authoritative public sources such as BSI (bsi.bund.de), EUR-Lex, and ENISA, which is consistent with its legal advisory purpose.
[PROMPT_INJECTION]: The skill processes user-provided documents, creating an interface for indirect prompt injection where untrusted data could influence the agent's behavior. However, the instructions include strict source-checking rules (Quellenregel) that require the agent to distinguish between model knowledge, user-provided text, and verifiable public sources, which mitigates standard injection risks.
Ingestion points: SKILL.md explicitly instructs the agent to work from user-provided documents, contracts, and screenshots.
Boundary markers: No specific boundary markers (like XML tags) are defined in the instructions.
Capability inventory: The skill is restricted to text analysis and drafting; no tools for file-writing, network requests (beyond documentation reference), or command execution are invoked.
Sanitization: No explicit sanitization logic is provided, but the skill mandates the marking of assumptions and missing evidence.
[REMOTE_CODE_EXECUTION]: No package installations or remote script executions were found.

spezial-nis2-livequellen-und-rechtsprechungscheck