pipeline-analysis
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from CRM exports or connected APIs to build a dashboard. This creates a surface for indirect prompt injection if the source data contains malicious instructions or scripts.\n
- Ingestion points: CRM data from files, sheets, or connected APIs as described in Section 1.\n
- Boundary markers: Absent. The skill does not instruct the agent to use delimiters or specific safety prompts to isolate external data from instructions.\n
- Capability inventory: The agent is instructed to generate and write a self-contained HTML file (Section 4), which could execute embedded scripts.\n
- Sanitization: Not specified. There is no instruction to escape or validate strings like deal names or account names before they are written to the HTML dashboard, which could lead to XSS if a user opens a dashboard containing malicious data.\n- [EXTERNAL_DOWNLOADS]: The dashboard generation process includes a reference to a third-party library for data visualization.\n
- Evidence: Section 4 specifies that using "Chart.js from a CDN is fine." This is a standard and safe practice for web-based dashboards using well-known services.
Audit Metadata