Hey Product Hunt! 👋 I'm Fred, co-founder of Plango. AI coding agents are incredible. They can read your code, run commands, install packages, access files, and communicate with the internet. That's also what makes them terrifying from a security perspective. We're giving AI agents increasingly powerful access to our machines, but very little oversight into what they're actually doing. **So Rich and I built Plango. 🛡️** Plango is a native macOS security suite built for developers using AI coding tools like Claude Code, Cursor, and Codex. It runs a local AI security model that evaluates threats around three things: **data access, communication, and exposure** — the lethal trifecta of agent security. Here's what Plango can do: 🔒 **Control agent access** — Restrict access to sensitive files, SSH keys, .env files, or even files marked with specific Finder color tags. 🌐 **Stop data exfiltration** — Monitor network activity, create block/allow lists, and control where apps and agents can communicate. 🔑 **Protect your secrets** — Move environment secrets into macOS Keychain. 🕵️ **Audit your Mac** — Review app permissions, privacy settings, startup items, and extensions. 🧩 **Scan skills and plugins** — Check agent skills and plugins for suspicious behavior. 🛠️ **Find vulnerabilities** — Scan projects across your Mac for security issues. 🤖 **Ask your security AI** — Investigate findings using a built-in AI assistant that runs locally, with options to use another local agent or your own API key. 📊 **See everything** — A dashboard of security events, requests, decisions, and blocked activity, with custom rules to make Plango work your way. **A little about us:** My co-founder Rich and I have a combined two decades of experience across cybersecurity, NLP, and AI. We're building Plango full-time because we believe AI-powered development is the future, while security and privacy must have a seat at the table. Our philosophy is simple: **Give agents the freedom to build, not to destroy.** Plango is available to try for free. We'd genuinely love your feedback, especially from anyone using AI coding agents daily. Given how fast cyber is evolving we're pushing changes to this on a daily basis. What worries you most about giving AI agents access to your machine? Fred & Rich ✌️

Plango

Maker

📌

11h ago

Mailwarm

moving the .env secrets into the keychain is the bit i'd turn on first, we have way too many of those lying around in old repos

does the local model jsut flag a weird outbound call or block it before it leaves??? curious how noisy it gets during npm installs

8m ago

SonOf

Using Finder tags to mark files the agent can't touch is a practical detail. That makes the boundary visible in a place where people already handle their files.

20m ago

Plango

Use AI to defend against AI

Use AI to defend against AI