workflow-site-audit
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources including raw HTML from target homepages (Phase 0), sitemap XML files, and third-party API results. This content is passed to multiple specialist subagents for analysis.
- Ingestion points:
Phase 0(webfetch of homepage, sitemap fetching),Phase 1(DataForSEO and Google API results). - Boundary markers: None specified; raw content is delegated directly to subagents.
- Capability inventory: The skill possesses extensive command execution capabilities (running numerous Python scripts) and file system write access for report generation.
- Sanitization: No explicit sanitization or instruction-ignoring delimiters are mentioned for the external content before it enters the agent's context.
- [COMMAND_EXECUTION]: The workflow is built around the execution of approximately eleven different Python scripts located in the
scripts/directory via shell commands. While these appear to be internal vendor-provided tools for SEO analysis, they provide the agent with broad capabilities to execute logic that processes user-supplied domains and external web data. - [EXTERNAL_DOWNLOADS]: The skill performs legitimate network operations to fetch data from the DataForSEO API, Google's PageSpeed and CrUX APIs, and the target website being audited. These operations are core to the skill's functional purpose.
Audit Metadata