The source shows a real, user-activated exposure of agent routes combined with default headless tool auto-approval. This is a dangerous capability warranting a warning, but the inspected files do not establish a malicious install-time action or a concrete attack independent of running the server.
Static reason
No blocking static signals were detected.
Trigger
A user or service starts the server or daemon in a process without a TTY, with CORTEX_HEADLESS_APPROVE unset or not false.
Impact
Requests reaching the exposed service may invoke agent tools without an interactive approval prompt; the inspected source does not establish what network boundary protects the listener.
Mechanism
The server binds to 0.0.0.0, mounts message and approval routes, and configures the orchestrator for automatic permission mode under the headless default.
Evidence
bin/cortex-daemon.jsdist/index.js