phishing-vorfall-pruefen
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, which creates a surface for indirect prompt injection where instructions hidden in bank documents or screenshots could influence the agent's behavior.
- Ingestion points: Processes user-provided documents, screenshots, account statements, bank letters, and ZIP archives (referenced in
SKILL.mdunder 'Sofortmodus'). - Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' instructions when analyzing these external sources.
- Capability inventory: The agent is tasked with generating legal documents (Aufforderungsschreiben, Klagegerüst) and executes a local script (
phishing_case_gate.py). - Sanitization: Absent. There are no instructions to validate or filter content extracted from external files before processing.
- [COMMAND_EXECUTION]: The skill provides instructions and examples for executing a local Python utility script.
- Evidence:
SKILL.mdprovides the commandpython phishing-vorfall-pruefen/scripts/phishing_case_gate.py --input testakten/phishing-vorfall-mayer-sparkasse-berlin/08_case_gate_input.jsonto process specific case files.
Audit Metadata