marketplace-personalisation
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides authoritative engineering guidance and reference implementations for AWS Personalize, drawing from official documentation and established industry blogs. All external references target well-known or trusted services and organizations.
- [EXTERNAL_DOWNLOADS]: The skill's setup instructions include using the pnpm package manager to install dependencies from the standard public registry.
- [INDIRECT_PROMPT_INJECTION]: The skill documents techniques for ingesting untrusted user inputs—such as onboarding preferences and explicit negative feedback—into recommendation pipelines, which represents a potential data ingestion surface for the final system.
- Ingestion points: Capture of user intent during registration (cold-capture-onboarding-intent.md) and seeker feedback events (track-capture-negative-signals.md).
- Boundary markers: Not explicitly detailed in the provided architectural examples.
- Capability inventory: The skill includes snippets for invoking AWS Personalize APIs (put_events, put_items, put_users) to store captured data.
- Sanitization: Code examples show direct ingestion of data objects; implementers should apply standard input validation and sanitization at the application layer.
Audit Metadata