Your AI agent no longer has to sit inside your terminal. Pi integration is now available in AI DevKit. Start a Pi session in Agent Console, or remote control it from Telegram with one click.

in the progress of testing Pi integration with AI DevKit, are you looking forward?

I used to copy messages between Codex and Claude like a human API. So I added a new agent skill for cross-agent session communication. Now Codex can talk to Claude, and Claude can talk back.

My goal when buidling AI DevKit also relate to this concept. Building the loop is designing the system that prompts the agents to do things instead of yourself. Your focus is on enhancing that system to ensure the output match your expectation and incline with your direction. The loop will require: - Development work can be paralleled, so that agent won’t step on each other feet, this is the concept behind AI DevKit dev-lifecycle skill - The work needs to be ready to handoff to another agent, dev-lifecycle artifact is solving this - Agent need to be able to observe and communicate with other agents, no matter what type (Claude, Codex, OpenCode, etc), to ask for info, delegate, etc, this is the motivation behind the AI DevKit agent manager - You will need a way to monitor what they (agents) are working on, this is the motivation behind the agent console - Agents need to share the same set of memory and easy to retrieve but not bloat the context, this is the motivation behind AI DevKit memory Of course this is not perfect and still very early state but I align with the ideas in overall, looking forward