The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, which represents a potential attack surface.\n- Ingestion points: The skill instructions specify collecting and analyzing external materials such as 'Verwaltungsakte, Vertragsurkunden, Schriftsätze, Bescheide, Protokolle, Sachverständigengutachten' and 'Screenshots oder Nachrichten' (SKILL.md).\n- Boundary markers: There are no explicit instructions to the agent to use delimiters or to ignore potential instructions embedded within the analyzed documents.\n- Capability inventory: The skill is restricted to generating text-based outputs (memos, checklists, tables, drafts). No dangerous capabilities like command execution, file system writes, or network requests were found in the skill logic.\n- Sanitization: No sanitization or validation logic for the external content is defined.

kanzlei-beweislast-und-darlegungslast