agent-manager-skill

Warn

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to download its functional components from a remote source at https://github.com/fractalmind-ai/agent-manager-skill.git.
  • [COMMAND_EXECUTION]: The skill executes local Python scripts (python3 agent-manager/scripts/main.py) that are part of the externally downloaded repository.
  • [PROMPT_INJECTION]: The assign command accepts multi-line task descriptions via shell heredocs. This serves as an ingestion point for untrusted data that could influence the behavior of the managed agents.
  • Ingestion points: python3 agent-manager/scripts/main.py assign command.
  • Boundary markers: Heredoc <<'EOF' is used, providing basic structure but not full protection against malicious instructions.
  • Capability inventory: Manages tmux sessions, starts/stops processes, and tails log files.
  • Sanitization: No sanitization or validation of the assigned task content is documented.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 20, 2026, 09:54 AM
Security Audit — agent-trust-hub — agent-manager-skill