amazon-review-strategy
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted user input to provide tailored strategic advice. While this represents a theoretical attack surface for prompt injection, the risk is negligible as the skill lacks any dangerous capabilities (e.g., file system access, network operations, or shell execution) to exploit.
- Ingestion points: Processes user-supplied product, platform, and goal information as described in Step 1 of the skill's workflow.
- Boundary markers: The instructions do not specify any delimiters or safety markers for processing this user data.
- Capability inventory: None. The skill does not perform any subprocess calls, file writes, or network requests.
- Sanitization: No explicit sanitization or validation of the user input is described.
- [NO_CODE]: The skill consists entirely of instructional Markdown text and lacks any scripts, binary files, or configuration files that could execute code on the host environment.
- [SAFE]: All strategies described (Amazon Vine, Request a Review, insert cards) are discussed within the context of compliance with Amazon's official terms of service. There are no findings related to obfuscation, credential theft, or unauthorized persistence.
Audit Metadata