akteneinsicht-auswerten
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructs the agent to manage sensitive client data within a specific local directory structure located in the user's home folder at
~/.claude/plugins/config/claude-fuer-deutsches-recht/sozialrecht-kanzlei/mandate/<az>/. This involves accessing and organizing information within potentially sensitive local file system paths. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret external, untrusted data from legal case files, including PDFs and OCR-scanned documents.
- Ingestion points: The skill takes 'Vollständige Verwaltungsakte' (PDF/OCR) and previous analysis protocols as input (SKILL.md).
- Boundary markers: There are no instructions for the agent to use specific delimiters or to disregard instructions that might be embedded within the processed legal documents.
- Capability inventory: The agent evaluates evidence, assesses contradictions, and generates legal draft components based on the ingested content.
- Sanitization: The process does not describe any validation or sanitization of content extracted from the external documents before it influences the agent's output.
Audit Metadata