The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, creating a potential surface for indirect prompt injection.
Ingestion points: The skill instructs the agent to analyze external "Dokumente und Beweismittel" (documents and evidence), "Sachverhalt" (facts), "Vertragsurkunden" (contractual documents), and "Sachverständigengutachten" (expert opinions) provided by the user.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands embedded within analyzed legal documents.
Capability inventory: No tools are explicitly requested in the allowed-tools frontmatter, limiting the potential impact of an injection.
Sanitization: There are no explicit sanitization or filtering instructions for the ingested content, though the skill requires verification of all claims against official sources.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch and verify information from official and well-known legal repositories.
Evidence: The instructions reference retrieving data from gesetze-im-internet.de, dejure.org, openJur, official BVerfG/BGH/EuGH databases, and guidance from the USPTO and EPO.
Context: These references are used for cross-referencing legal norms and jurisprudence, which is consistent with the skill's stated purpose.

patent-infringement-software