Hi Product Hunt! 👋 I built CodePet after noticing a strange side effect of using more AI coding tools: the agents became more capable, but the project context became more fragmented. A task might start in Codex, continue in Cursor or TRAE, and later be handed to a teammate. The code is still there, but the goal, decisions, validation results, risks, and next step are often trapped in separate chats or in one person's memory. CodePet is not another coding agent. It is a local-first desktop companion for the agents you already use. It helps you: 1、notice key task states without watching every window; 2、create checkpoints before changes and keep validation separate from completion; 3、record structured progress—goal, completed work, open items, decisions, artifacts, validation, risks, and next step; 4、preview and edit a handoff pack before sharing it with another agent or teammate. We deliberately keep the evidence boundaries visible: copying a handoff does not mean it was delivered, delivery does not mean it was executed, and task completion does not mean validation passed. When a platform does not provide an open handoff API, CodePet uses an explicit manual copy-and-paste flow rather than pretending the transfer happened automatically. The cross-agent handoff flow has been validated from Codex to TRAE. We are now collecting real-team feedback before making any productivity or token-saving claims. I'd love your feedback: which agents do you switch between, and what context do you most often lose during a handoff? Learn more and download CodePet at https://codepet.me

CodePet

Keep AI coding moving across agents and teammates

Keep AI coding moving across agents and teammates