qonto-tax-pilot
Warn
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: MEDIUMINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external transaction and invoice records which may contain malicious instructions designed to influence the agent's behavior. \n
- Ingestion points:
list_transactions,list_client_invoices, andlist_supplier_invoices(SKILL.md). \n - Boundary markers: Absent; there are no specific delimiters or instructions provided to the agent to disregard content within the data fields. \n
- Capability inventory: The skill has the ability to prepare multi-transfer requests and generate executable UI components (HTML). \n
- Sanitization: No sanitization or validation logic is defined to filter transaction descriptions or invoice metadata before they are processed. \n- [DYNAMIC_EXECUTION]: The skill generates interactive HTML content at runtime based on ingested external data. \n
- Evidence: The
Reportsection ofSKILL.mddescribes the generation of an interactive HTML dashboard incorporating data from transactions and invoices. This creates a risk of Cross-Site Scripting (XSS) if the host environment renders the generated code without sufficient escaping of the counterparty names or invoice notes.
Audit Metadata