twilio-email-send
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data for the purpose of sending emails, including recipient addresses, subject lines, body content, and Liquid template variables. This creates a surface where malicious instructions embedded in data from untrusted sources (such as a website being summarized or an incoming message) could potentially influence the agent's behavior or be used to send unauthorized content.
- Ingestion points: The
torecipient array,content.subject,content.html,content.text, and thevariablesobject referenced inSKILL.mdare all entry points for external data. - Boundary markers: The
Overviewsection provides a clear safety instruction: "Always confirm recipients, subject, and content with the user before sending. Email is irreversible once delivered. Never send email autonomously without explicit user approval." - Capability inventory: The skill utilizes network capabilities to perform POST requests to
https://comms.twilio.com/v1/Emails. - Sanitization: The skill documentation does not specify automated sanitization or filtering of the inputs, relying instead on the human-in-the-loop confirmation requirement.
Audit Metadata