seo-report
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted research data and existing report content (via
get_report) to generate new HTML reports. This data could contain malicious instructions designed to influence the report's content or manipulate the agent's behavior during the saving and sharing process. - Ingestion points: The skill reads research data from preceding skill runs, report summaries, and full HTML reports via the
get_reporttool. - Boundary markers: The instructions do not specify explicit delimiters or "ignore" instructions for the research data being interpolated into the report template.
- Capability inventory: The agent can save files (
save_report) and modify public visibility of reports (set_report_sharing). - Sanitization: The skill enforces strict HTML security constraints—such as forbidding external resources (web fonts, CDNs,
fetch) and executable<script>tags—which effectively mitigates many downstream exploitation vectors in the generated report, though it does not prevent the agent itself from being influenced by the input data.
Audit Metadata