cmux-workspace

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the cmux CLI to perform environment mutations, including creating panes (new-pane), sending keyboard input (send-key), and managing workspaces.
  • [COMMAND_EXECUTION]: It instructs the agent to execute local scripts (./scripts/reload.sh and ios/scripts/reload.sh) found within the user's working directory to perform environment reloads.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by reading terminal output into the agent's context using the cmux read-screen command.
  • Ingestion points: cmux read-screen (used in SKILL.md to verify terminal state).
  • Boundary markers: Absent; there are no instructions to delimit terminal content or ignore embedded commands within the ingested data.
  • Capability inventory: The agent has capabilities to execute shell commands (cmux send), run local scripts (reload.sh), and open files (cmux open).
  • Sanitization: The skill does not specify any sanitization, filtering, or validation for the content read from the terminal screen before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 07:18 AM
Security Audit — agent-trust-hub — cmux-workspace