market-audit
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external content fetched from user-provided URLs, creating an ingestion surface where third-party web content could contain embedded instructions designed to influence the subagents' analysis or modify the audit outcomes.
- Ingestion Points: Ingests untrusted external data in Phase 1 via
python3 scripts/analyze_page.py <url>insideSKILL.md. - Boundary Markers: Absent. The resulting JSON payload is fed directly into multiple subagents without explicit structural isolation or boundary enforcement rules.
- Capability Inventory: Executes
python3 scripts/analyze_page.pyand writes output toMARKETING-AUDIT.md. - Sanitization: Absent. No filtering, input sanitization, or validation constraints are applied to the fetched webpage data before it is propagated across the subagent pipeline.
Audit Metadata