service-innovation-case-study
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a methodological approach to business research, focusing on information organization and reporting. It contains no instructions for privilege escalation, data exfiltration, or persistence.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through external research, a standard vector for indirect prompt injection.
- Ingestion points: External sources such as LinkedIn, App Store reviews, and industry databases are processed to build reports.
- Boundary markers: The 'Research Protocol' (references/01-research-protocol.md) enforces boundaries by requiring the agent to distinguish between company claims and independent evidence.
- Capability inventory: The skill's primary function is text generation; it does not possess capabilities for arbitrary command execution or network exfiltration.
- Sanitization: The methodology requires source validation and the explicit documentation of information gaps, reducing the risk of accepting malicious content as authoritative.
Audit Metadata