skills/klotzkette/claude-fuer-deutsches-recht/rechtsabteilung-us-work-made-for-hire-im-deutschen-projekt/Gen Agent Trust Hub
rechtsabteilung-us-work-made-for-hire-im-deutschen-projekt
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions define a workflow for ingesting and processing external, potentially untrusted data such as emails, logs, and customer communication.
- Ingestion points: The 'Sofortprüfung' section directs the agent to secure and analyze 'E-Mails', 'Logs', and 'Kundenkommunikation'.
- Boundary markers: The skill does not currently provide explicit delimiters or instructions for the agent to separate embedded instructions from document data.
- Capability inventory: The agent uses standard text processing and summarization capabilities to produce legal memos.
- Sanitization: No specific filtering or sanitization steps are included for the data ingestion process.
- [EXTERNAL_DOWNLOADS]: The skill references official legal information portals for the verification of laws and regulations.
- The instructions recommend checking sources such as 'gesetze-im-internet.de', 'dejure.org', and 'openJur' for legal citations.
- [NO_CODE]: The analyzed skill consists entirely of markdown-based instructional content and does not include any executable scripts, binary files, or automated configuration tools.
Audit Metadata