Your AI coding assistant isn't a chatbot anymore. It clones repos, installs packages, runs commands, and wires up tools — all with your credentials. You'd never give a brand-new intern root access on day one… yet by default, that's exactly what we hand our agents.

In 45 fast, story-driven minutes we follow how real attacks actually happen: the npm heist that poisoned libraries with two billion weekly downloads, invisible malware hiding in VS Code extensions, prompt injection that quietly hijacks your agent, and booby-trapped MCP servers and CLIs. No jargon, no fear-mongering — every cinematic attack is paired with the dull, one-line "chore" that beats it: lockfiles, pinning, fencing, scanning, gates.

You'll leave able to vet a repo, a package, and an MCP before you run them, keep your agent from acting on manipulative instructions, and treat security as a habit, not a firewall. Live demos included.

Your AI coding assistant isn't a chatbot — it's a developer with tools, running commands with your credentials. This fast, story-driven session shows how attackers exploit that — supply chain, prompt injection, MCP & CLI — and the boring habits that stop them. Real breaches, live demos, zero fluff.