lieferbedingungen-adsp
Pass
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted data in the form of contract details, party positions, and case facts to generate legal opinions. This represents an indirect prompt injection surface where instructions could theoretically be embedded in the analyzed legal text to influence the agent's behavior.
- Ingestion points: User-provided inputs for 'Vertragstyp', 'Position des Mandanten', and 'Konkrete Streitklausel' defined in SKILL.md.
- Boundary markers: Not explicitly implemented in the prompt instructions.
- Capability inventory: The skill is limited to legal reasoning and document drafting; no dangerous file system or network execution tools are identified.
- Sanitization: No specific filtering or sanitization of input legal text is described.
- [EXTERNAL_DOWNLOADS]: The skill references several external URLs to fetch legal statutes and financial data. All identified links target official government repositories or well-known international organizations, including the German Federal Ministry of Justice (gesetze-im-internet.de), the International Monetary Fund (imf.org), and the DSLV industry association (dslv.org). These are legitimate references for the skill's legal purpose and do not involve the execution of remote code.
Audit Metadata