codex-voice-optimizer

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of 'MSL' (Minimum Sufficient Language) and 'MSW' (Minimum Sufficient Work) companion kernels from the 'transcendr/slopware-skills' marketplace. These are described as optional enhancements and require explicit user authorization before the installation commands are executed.
  • [COMMAND_EXECUTION]: The skill uses specific application tools to manage lifecycle and tasks within the Codex environment. These include codex_app__list_projects, codex_app__create_thread, and codex plugin add. These commands are constrained to the application's internal plugin and task management systems.
  • [PROMPT_INJECTION]: The skill includes logic to prioritize its own orchestration instructions over general model behavior during voice sessions, including rules to ignore irrelevant background audio and strictly follow role contracts. These are functional constraints designed for voice interface reliability.
  • [DATA_EXPOSURE]: The skill uses a native inline route (::codex-realtime-inline{}) to render local file paths and artifacts in the chat interface. This is used for project navigation and plan visibility within the desktop application environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 04:15 PM
Security Audit — agent-trust-hub — codex-voice-optimizer