task-supervisor

Pass

Audited by Gen Agent Trust Hub on Apr 4, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's logic is entirely consistent with its described function as a supervisor for monitoring and managing agent tasks.
  • [DATA_EXPOSURE]: The skill collects metadata from other active sessions (such as activity timestamps and stop reasons) and transmits reports to a hardcoded target ID (-1003890797239). This behavior is a functional requirement for the skill's alerting and reporting system.
  • [PROMPT_INJECTION]: The skill monitors data from other agent sessions, which creates a surface for indirect prompt injection.
  • Ingestion points: Session metadata and message statuses are accessed via the sessions_list tool in SKILL.md.
  • Boundary markers: No specific delimiters or "ignore instructions" markers are used when evaluating session content.
  • Capability inventory: The skill uses the sessions_send tool to send messages to other sessions and perform automated task restarts.
  • Sanitization: No explicit sanitization or validation of the monitored session data is implemented.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 4, 2026, 11:47 AM