The Safest Way I've Found to Let AI Coding Agents Run Real Infrastructure

תקציר מקורי באנגליתI've been using AI coding assistants to write all of my code this entire year! I still validate everything though, and what limits my trust isn't the model - it's what it can reach. The second an agent touches real infrastructure you're handing it shell access, a database URL and a cloud key. Now it can do anything, including wiping your database. Kestra does it the other way around. The operation is YAML in Git, and the agent gets a list of flows it's allowed to run instead of the credentials. Every call comes back as a normal execution you can open and read. Kestra 2.0 is officially out now, and the core stays Apache 2.0. Kestra 2.0: https://fandf.co/4gEtWhm Clone this AI agent blueprint to try it: https://fandf.co/3VcvAQf GitHub: https://fandf.co/4h8rlh9 Thanks to Kestra for working wit