The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze materials and documents that may originate from untrusted sources, creating an attack surface for embedded instructions.
Ingestion points: The agent is directed to read provided materials and collected evidence such as contracts, administrative acts, protocols, and expert opinions as described in the 'Arbeitsweg' section.
Boundary markers: The instructions do not define specific delimiters or guidelines to distinguish between data and potential commands within the analyzed documents.
Capability inventory: The agent's tasks involve textual analysis, summarization, and generating structured matrices and protocols. It does not utilize high-risk tools for arbitrary code execution or unauthorized network access.
Sanitization: There are no explicit instructions for the agent to sanitize, escape, or filter content from external materials before processing.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to verify information against authoritative legal databases.
The instructions mandate checking legal norms and court decisions via established services such as gesetze-im-internet.de, dejure.org, openJur, and official databases for BVerfG, BGH, and EuGH.
[NO_CODE]: The skill consists of instructional text in markdown format and does not include any executable scripts or binary components.

dsv-beweissicherung