content-platforms
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides interfaces and functions for building CMS platforms that process untrusted user-generated content, creating a risk of indirect prompt injection if an agent processes this data without sufficient isolation. \n
- Ingestion points: Content data processed in
createVersion(SKILL.md), rich text HTML output from theRichTextEditorcomponent (SKILL.md), and file metadata in theuploadMediafunction (SKILL.md). \n - Boundary markers: None present; the templates provide structural logic but do not include prompt delimiters or instructions for the agent to disregard embedded content. \n
- Capability inventory: The skill utilizes capabilities for cloud storage writes (AWS S3
PutObjectCommand) and database interactions (Prismacreateandupdatecalls). \n - Sanitization: The provided boilerplate code extracts HTML and metadata from inputs without explicit sanitization or filtering logic shown in the examples.
Audit Metadata