Claude Code is one capable coding agent in your terminal; o8 runs it — alongside Codex, Gemini, and opencode — as agents in a governed fleet. o8 is the room; Claude Code is a runtime in it.
Claude Codeis Anthropic’s agentic coding tool: a single, very capable agent that works in your terminal, reads and edits your repo, and runs commands. It’s one agent at your prompt, doing one stream of work.
o8 is a local-first macOS control room for autonomous engineering. One orchestrator dispatches a fleet of coding agents — each in its own isolated git worktree — and a governance layer reviews every diff. Nothing merges unreviewed, and nothing risky merges without your sign-off. It runs on your own machine, against your own model keys, and works with Claude, Codex, Gemini, and opencode through one adapter.
o8 doesn’t replace Claude Code — it orchestrates it. Claude Code (and Codex, Gemini, opencode) are the runtimes o8 dispatches work to. o8 adds the layer above a single agent: many of them running in parallel in sealed worktrees, a review gate between their diffs and your main branch, conflict detection, audit, and memory that turns repeated fixes into proposed rules. One agent becomes a fleet you can govern.
Use Claude Code on its own for a single focused task in the terminal; reach for o8 when you want several agents — including Claude Code — running at once with review and audit around them. See the docs.