velt-notifications-best-practices
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents how to create custom notifications that interpolate user-supplied data into message templates, which represents a vulnerability surface for indirect prompt injection if the resulting notifications are processed by an AI agent.
- Ingestion points: The
displayHeadlineMessageTemplateDataanddisplayBodyMessagefields inrules/shared/triggers/triggers-custom.mdallow for external data ingestion. - Boundary markers: The implementation examples provided do not include specific boundary markers or 'ignore' instructions for the data being interpolated.
- Capability inventory: As a documentation skill, it does not possess direct execution capabilities but describes the construction of systems with network (REST API) and UI capabilities.
- Sanitization: The patterns do not explicitly mention or demonstrate the sanitization of external data before it is included in notification headlines or bodies.
- [SAFE]: The skill references official vendor domains (velt.dev, console.velt.dev) and packages (@veltdev/react) that are consistent with the author 'velt-js'.
- [SAFE]: Hardcoded string values such as 'YOUR_API_KEY' and 'YOUR_AUTH_TOKEN' are clearly marked documentation placeholders and do not contain real credentials.
Audit Metadata