sinch-functions-node
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses official vendor resources from 'sinch', including the '@sinch/functions-runtime' package and 'developers.sinch.com' domain.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies ingestion of untrusted data from voice and conversation webhooks. It explicitly instructs the agent to treat this data as malicious and avoid unsafe interpolation, mitigating the risk of injection attacks.
- Ingestion points: Inbound call events and conversation message payloads mentioned in
SKILL.mdandreferences/conversation-webhooks.md. - Boundary markers: Present; the skill specifies that callback data is untrusted and provides a 'Security' section with explicit constraints.
- Capability inventory: Access to Sinch SDK services (Voice, SMS, Conversation), persistent storage, and SQLite database operations.
- Sanitization: Instructions mandate input validation and forbid interpolation into prompts or shell commands.
Audit Metadata