skills/thaitype/skills/claude-tui/Gen Agent Trust Hub

claude-tui

Warn

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/tui_cmd.py is vulnerable to shell command injection when initializing a tmux session.
  • Evidence: In the run_tui_command function, the cwd variable (derived from a session's JSON data or the --dir argument) is interpolated directly into a shell command string: f"cd {cwd} && claude --resume {session_id}".
  • Risk: A maliciously crafted session file or an untrusted path passed to --dir could contain shell metacharacters (e.g., ;, &, |) resulting in arbitrary command execution when tmux new-session is called. While the Python subprocess.run call uses a list for the outer tmux command, the inner command passed to tmux's shell remains unsanitized.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 20, 2026, 10:47 AM
Security Audit — agent-trust-hub — claude-tui