The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could potentially be used as a vector for indirect prompt injection.
Ingestion points: The skill is instructed to request and analyze user-provided data including original files, timestamps, versions, metadata, contracts, screenshots, and platform notifications (found in SKILL.md under 'Sofortstart').
Boundary markers: The instructions lack explicit delimiters or safety markers to differentiate between user-provided data and the agent's core operating instructions.
Capability inventory: The skill has the capability to perform live web checks and access external legal databases to verify norms and case law.
Sanitization: There is no evidence of sanitization, validation, or filtering of the content extracted from the ingested files.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch legal data from several external services.
Evidence: The agent is directed to perform live verification of legal norms using domains such as gesetze-im-internet.de, dejure.org, eur-lex.europa.eu, and openJur (found in SKILL.md under 'Arbeitsweg').
Context: These are well-known, authoritative legal repositories. The network access is used to ensure the agent uses up-to-date legal information rather than relying solely on training data, which aligns with the skill's stated legal purpose.

verjaehrung-fristen