@prestyj/boss@5.4.0
Orchestrator agent that drives multiple ezcoder sessions across projects from a single chat
AI security review
scanned · by lpm-firewall-aiStatic source inspection shows no lifecycle hook, import-time payload, credential harvesting, or unconsented mutation of a foreign/broad AI-agent control surface. The scanner's exfiltration and shell findings map to explicit CLI features, bundled pixel helper code, Telegram setup, radio playback, worker tooling, and self-update behavior rather than a concrete malicious chain.