review-response
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted 'Review text' provided by the user or retrieved from external platforms to generate responses. An attacker could embed malicious instructions within a review to influence the agent's behavior.
- Ingestion points: Untrusted 'Review text' and metadata (rating, platform) enter the context via SKILL.md. The skill also reads brand-specific files from
~/.claude-marketing/. - Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when handling the external review content.
- Capability inventory: The skill's capabilities are limited to text analysis, drafting, and scoring. It does not use tools for network access, file system modification, or command execution.
- Sanitization: No sanitization, filtering, or validation steps are mentioned for the external content before it is passed to the LLM agents.
Audit Metadata