Claude Code, Codex, and Grok, side by side — started, watched, and approved exactly the same way. Plus any agent that speaks the open Agent Client Protocol.
Choose right from the model picker — a Claude model runs Claude Code, a GPT or Codex model runs Codex, a Grok model runs Grok. Run them at the same time, on the same project, and compare their work in one review surface.
Nucleic speaks the Agent Client Protocol — an open, standard way for editors and agents to talk. Any ACP agent joins your fleet behind the same approvals and review. The newest capable agent can slot in without changing how you work.
Some tasks suit one agent better — pick per task instead of committing to one.
Your workflow, approvals, and review stay the same as agents come and go.
Each agent authenticates with your credentials — nothing sits between you and your providers.
Same statuses, same approvals, same merge flow, whichever agent is working.
Nucleic is in open beta. All it asks is macOS 27 on Apple Silicon.