Plango: On‑Device AI Security for Mac AI Coding Agents

Plango is a local AI security model that runs directly on Apple‑silicon Macs (macOS 15 or newer) to monitor AI coding agents, protect sensitive files, and block suspicious network activity. It ships as a trial with no purchase required and integrates with common AI development tools such as Claude Code, Cursor, GitHub Copilot, Zed, and the OpenAI Codex CLI. Setup basics Installation starts from the product website; the installer places a background service that watches agent launches and network sockets. Because the model is on‑device, no external API keys are sent to third parties, which reduces latency and eliminates data‑exfiltration risk through the cloud. First‑time users should verify two prerequisites: an Apple‑silicon Mac and macOS 15+. If either condition fails, Plango will prompt you to upgrade or switch hardware before any protection can be active. Key controls and failure modes Plango offers the following mechanisms to harden an AI‑driven workflow: Agent‑access fencing – SSH keys, .env files, and Finder‑tagged folders are hidden from agents. Network‑activity watch – every outbound connection is logged and can be allowed or denied. Secret migration – plain‑text credentials are moved to the macOS Keychain. Skill/plugin scanning – hidden instructions in extensions or custom plugins are flagged. Vulnerability detection – scans for