dpdpa
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted documents provided by the user for compliance analysis, which presents a surface for indirect prompt injection.\n
- Ingestion points: Untrusted data enters the context when users provide draft notices, privacy policies, or vendor contracts for review, as described in the 'How to Respond' section of SKILL.md.\n
- Boundary markers: There are no instructions in SKILL.md that tell the agent to ignore or delimit instructions potentially embedded within the documents being reviewed.\n
- Capability inventory: The skill's capabilities are limited to text generation; a review of SKILL.md and all reference files (gdpr-comparison.md, rights-and-obligations.md, rules-2025.md, sections-reference.md) shows no file-system write operations, network calls, or subprocess executions.\n
- Sanitization: No input sanitization or filtering mechanisms are described in the skill instructions.\n- [NO_CODE]: The skill is entirely composed of informational markdown files and does not include any executable scripts, dependencies, or system configurations.
Audit Metadata