twilio-sms

Warn

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Modifies shell profile configuration (~/.zshrc) to persist environment changes and path updates for Node.js and Python runtimes.
  • [COMMAND_EXECUTION]: Implements persistence for the messaging application by providing a template for a systemd service unit.
  • [COMMAND_EXECUTION]: Utilizes sudo for system-level operations, including repository management and the installation of core packages and the Twilio CLI.
  • [REMOTE_CODE_EXECUTION]: Downloads and executes an installation script from NodeSource using a piped bash command.
  • [PROMPT_INJECTION]: Processes untrusted data from inbound SMS bodies, which could be exploited for indirect prompt injection if the agent uses the content to drive further logic.
  • Ingestion points: Inbound SMS message body received via webhooks in the /twilio/inbound endpoint (demonstrated in Node.js and Python examples).
  • Boundary markers: The implementation lacks delimiters or instructions to ignore embedded commands within the message body.
  • Capability inventory: The skill can generate automated responses via TwiML, interact with external ticketing systems (as shown in pseudo-code), and perform bulk messaging operations.
  • Sanitization: Basic string normalization (trimming and case conversion) is present, but no robust sanitization or escaping is implemented for the external content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 7, 2026, 05:45 PM