kanzlei-automationen-bea-journal

Pass

Audited by Gen Agent Trust Hub on Sep 29, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines workflows for processing untrusted external data such as beA (Electronic Attorney Mailbox) messages, emails, and business documents which may contain malicious instructions designed to influence the agent.
  • Ingestion points: Data enters the context through folder monitoring, beA journal exports, email integration (Outlook), and document collection for tax/payroll routines.
  • Boundary markers: The skill explicitly instructs the agent to maintain transparency and requires human-in-the-loop verification for every dispatch, EB (Electronic Acknowledgement of Receipt), and message export.
  • Capability inventory: Capabilities include file system monitoring, ZIP archive extraction (Entpacken), and generating text for drafts, checklists, and calendar entries.
  • Sanitization: The skill provides strict prohibitions against accepting sensitive credentials (PINs, tokens, passwords) and prevents the automated submission of tax filings (UStVA) or bank transfers.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 29, 2026, 03:55 PM