gesellschaftsrecht
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [DATA_EXPOSURE]: The skill is configured to read from a local workspace path at
~/.config/claude-fuer-deutsches-recht/grosskanzlei-corporate-ma/mandate/<slug>/. This directory contains mandate-specific information such asmandat.md,history.md, andfristen.yaml. While this path is outside the immediate project root, it is utilized as the designated environment for mandate management within the specialized toolset. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes untrusted data from external legal documents, including SPA drafts, disclosure letters, and board approvals.
- Ingestion points: Files loaded from the mandate workspace and user-provided document uploads.
- Boundary markers: The instructions mandate the use of source-specific tagging (e.g.,
[Mandant],[Register],[BGH-Datenbank]) to distinguish external data from model knowledge. - Capability inventory: The skill focuses on analytical output and drafting; it includes explicit prohibitions against automatic external communication.
- Sanitization: Emphasizes the need for manual verification (
[zu verifizieren]) of register entries and court decisions, reducing the risk of accidental obedience to embedded instructions.
Audit Metadata