yahav-design-agent
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where the agent reads and processes untrusted content from external JSON files to generate React component code.
- Ingestion points: The skill instructions explicitly direct the agent to read context files from the filesystem, specifically
messages/[timestamp]/copy.jsonandmessages/[timestamp]/design-brief.json(as seen in SKILL.md). - Boundary markers: There are no instructions or delimiters provided to the agent to treat the content within these JSON files as data only or to mitigate potential instructions embedded within the text fields of the JSON.
- Capability inventory: The skill leverages the agent's ability to generate functional React/Next.js code and uses a 'Read' tool to fetch project-specific data from the message context.
- Sanitization: The skill lacks validation or sanitization rules for inputs before they are interpolated into the component templates found in
references/section-components.md.
Audit Metadata