build-gate-visual-review
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data, including user-pasted snippets and external references. It provides a robust security framework to mitigate injection risks by requiring the agent to filter, quote, and paraphrase content while explicitly stripping any embedded instructions.
- [COMMAND_EXECUTION]: The skill uses the
findcommand for a maintenance task to identify stale HTML files older than 30 days. This execution is non-automated and gated behind an explicit user confirmation prompt (Archive stale HTML? y/n). - [SAFE]: The skill defines strict 'Safe HTML constraints' for its output, requiring all source text to be escaped and forbidding the use of inline event handlers, unauthorized scripts, or external resource references.
- [SAFE]: Secure delegation boundaries are established for child agents, restricting their read/write access to specific paths, forbidding network operations, and preventing environment or configuration changes.
Audit Metadata