amazon-return-reduction
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists exclusively of Markdown documentation (SKILL.md and README_CN.md). There are no scripts, binaries, or configuration files that execute logic. The functionality is entirely prompt-driven and advisory in nature.
- [SAFE]: No network operations, file system modifications, or credential handling behaviors were detected. The skill operates as a static methodology for the agent to follow when interacting with the user.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data provided by the user, such as "Amazon Return Analysis Reports" and "Buyer Reviews" (SKILL.md, Step 1).
- Ingestion points: Data provided by the user during the conversational flow (SKILL.md, Data Sources section).
- Boundary markers: None identified; the skill does not use specific delimiters to wrap user-provided data.
- Capability inventory: No capabilities (file writes, network calls, or subprocess execution) are defined in the skill or requested in frontmatter.
- Sanitization: No explicit sanitization of user input is mentioned.
- Analysis: While the skill processes untrusted external content (reviews), the total lack of exploitable tools or capabilities mitigates the risk of indirect injection to a safe level.
Audit Metadata