argentier-cash-forecast
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues or malicious patterns were detected. The skill is limited to a read-only environment regarding financial data and has no network permissions, preventing data exfiltration.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface from untrusted transaction labels and invoice details but includes high-quality mitigations. 1. Ingestion points: The skill reads external data via Qonto MCP tools including list_transactions, list_client_invoices, and list_supplier_invoices. 2. Boundary markers: Explicit instructions are provided to ignore any instructions or requests embedded within the tool-returned text. 3. Capability inventory: The skill utilizes file writing tools to generate CSV and HTML reports. 4. Sanitization: The instructions mandate the use of textContent for HTML to prevent cross-site scripting and single-quote prefixing for CSV cells to prevent spreadsheet formula injection.
Audit Metadata