listener-creator

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it interpolates untrusted email content (subject and body) directly into prompts for the callAgent method (e.g., in ai-classifier.ts and urgent-watcher.ts). \n
  • Ingestion points: Untrusted data enters via the Email object fields (email.subject, email.body).\n
  • Boundary markers: No delimiters (like XML tags) or instructions to ignore embedded commands are present to separate the email content from the system instructions.\n
  • Capability inventory: The skill can archive, star, label emails, and send high-priority notifications, all of which could be manipulated by a malicious email.\n
  • Sanitization: The skill only performs basic string truncation on the email body, providing no protection against adversarial content.\n- [COMMAND_EXECUTION]: The skill's primary function is to generate and write TypeScript source code to the agent/custom_scripts/listeners/ directory. These files are then dynamically loaded and executed by the platform to monitor email events, representing a local code execution surface.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 04:52 PM
Security Audit — agent-trust-hub — listener-creator