The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of legal documents, evidence, and metadata provided by the user. This ingestion surface is susceptible to indirect prompt injection, where instructions hidden within the evidence could attempt to manipulate the agent's legal analysis or output.
Ingestion points: The skill processes user-provided documents including administrative acts, contracts, court records, and screenshots as described in SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are defined for the input data in SKILL.md.
Capability inventory: The skill generates structured legal analysis (risk matrices, memos, and calculation sheets) but does not use dangerous capabilities like shell command execution or file writing.
Sanitization: The instructions do not define methods for sanitizing or filtering the content of the ingested evidence.
[EXTERNAL_DOWNLOADS]: The instructions require the agent to fetch and verify legal norms and case law from external sources.
Evidence: The skill references official and well-known legal resources including gesetze-im-internet.de, dejure.org, eur-lex.europa.eu, openJur, and the databases of the BVerfG, BGH, and EuGH.

schadensersatz-lizenzanalogie