audience-clarity-review
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary HTML content and CONTEXT.md files, creating an attack surface where instructions embedded in the data could influence the sub-agents. Ingestion points: HTML and CONTEXT.md files are read in Step 1 and passed to sub-agents in Step 5. Boundary markers: No delimiters or ignore instructions are used to isolate untrusted data. Capability inventory: Sub-agents write findings to the filesystem and return data via stdout. Sanitization: No sanitization is performed on external content.
- [COMMAND_EXECUTION]: The skill executes a local Node.js script to perform browser automation.
- [PRIVILEGE_ESCALATION]: The screenshot script uses the --no-sandbox flag for Chromium, which is a security reduction in browser isolation.
- [DYNAMIC_EXECUTION]: The script uses dynamic path resolution via environment variables to load the playwright library.
Audit Metadata