skills/1999azzar/tmux-skill/tmux/Gen Agent Trust Hub

tmux

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's primary function is to send arbitrary keystrokes and commands to interactive shells via tmux send-keys. This provides the agent with full terminal access within the tmux session.
  • [INDIRECT_PROMPT_INJECTION]: The skill explicitly instructs the agent to capture pane output (capture-pane) to monitor state and synchronize actions. If the terminal output contains data from external sources (e.g., logs, file contents, or HTTP responses), it could contain malicious instructions designed to subvert the agent's behavior.
  • [SAFE_PRACTICE]: The instructions mandate the use of a dedicated socket directory (AGENT_TMUX_SOCKET_DIR) to ensure the agent's environment is isolated from the user's personal tmux sessions, preventing accidental interference or data leakage between environments.
  • [PERSISTENCE]: By design, the skill utilizes tmux to maintain long-running stateful sessions. While this is a form of persistence, it is the stated purpose of the skill and is managed within isolated sockets.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 06:18 PM
Security Audit — agent-trust-hub — tmux