build-resource-pages
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to malicious payloads or instructions embedded in the data it processes.
- Ingestion points: The skill reads and processes markdown files from the
workspace/[brand]/content/resources/directory (SKILL.md). - Boundary markers: There are no instructions to use delimiters or ignore embedded instructions when parsing these external content files.
- Capability inventory: The skill identifies the frontend codebase, modifies existing web application code, and generates new routes and components that are intended for production use.
- Sanitization: The instructions specifically state to "Handle JSON-LD script blocks in the markdown (pass through to rendered output, don't strip)," which creates a direct vector for Cross-Site Scripting (XSS) if an attacker can influence the content files.
- [PROMPT_INJECTION]: The instructions include directives aimed at bypassing standard human-in-the-loop safety protocols.
- Evidence: The README.md and SKILL.md files explicitly state that "Output ships directly to production" and "No review pass needed." These instructions attempt to remove standard security constraints and oversight during the deployment of AI-generated code.
Audit Metadata