billing-automation

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill consists of functional markdown instructions and Python code snippets for billing logic. The content is educational and follows industry-standard practices for the described domain.
  • [PROMPT_INJECTION]: The skill defines logic for processing external customer data into automated templates, which is a standard surface for indirect prompt injection.
  • Ingestion points: Customer metadata (name, address, email) used in Invoice.to_html and DunningManager.send_dunning_email in resources/implementation-playbook.md.
  • Boundary markers: None implemented in the provided code snippets.
  • Capability inventory: Email transmission via send_email and file/document generation via to_pdf and to_html.
  • Sanitization: No specific escaping or validation logic is shown for the customer-supplied strings.
  • [DATA_EXFILTRATION]: No hardcoded credentials or unauthorized exfiltration patterns were found. The use of the Stripe API for payment processing is consistent with the skill's primary purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 01:56 PM
Security Audit — agent-trust-hub — billing-automation