clausona / site
llms.txt60 starsChanged 30 days agoMITView on GitHub
clausona is a free, open-source CLI and terminal dashboard for running multiple Claude Code and OpenAI Codex CLI accounts on one machine. Each account keeps its own sign-in in its own config directory (CLAUDE_CONFIG_DIR or CODEX_HOME). Skills, hooks, plugins, settings and global instructions are shared between the accounts of each tool, and conversation history can be shared too, so a conversation started under one account can be resumed under another. It also shows every account's 5-hour and weekly plan usage…
- Pipes a download into a shell
The file
Files
# clausona > clausona is a free, open-source CLI and terminal dashboard for running multiple Claude Code and OpenAI Codex CLI accounts on one machine. Each account keeps its own sign-in in its own config directory (CLAUDE_CONFIG_DIR or CODEX_HOME). Skills, hooks, plugins, settings and global instructions are shared between the accounts of each tool, and conversation history can be shared too, so a conversation started under one account can be resumed under another. It also shows every account's 5-hour and weekly plan usage side by side. It runs on macOS, Linux and Windows, has no proxy or daemon, and is not affiliated with Anthropic or OpenAI. Facts: - Install: `curl -fsSL https://github.com/larcane97/clausona/releases/latest/download/install.sh | bash` (macOS, Linux) or `irm https://github.com/larcane97/clausona/releases/latest/download/install.ps1 | iex` (Windows PowerShell). Needs Node.js 20 or later. - Set up: `clausona init` registers the accounts already signed in; `clausona add claude:work` or `clausona add codex:work` adds one; `csn use work` switches; `clausona run claude:personal` runs one session under another account. - Shared history: `clausona config <profile> --merge-sessions`; then `claude --resume` and `claude --continue`, and on macOS and Linux `codex resume`, see conversations from every account that shares history. - Never shared: sign-ins, Claude Code's `.claude.json` (including `claude mcp add` servers) and the state it keeps for one account or org (policy limits, remote managed settings, the MCP sign-in cache, the background daemon with its scheduled tasks and Remote Control folders, which a profile sets up on its own), and Codex's auth.json, app-server daemon, memories and its databases other than the thread store (which merged sessions share on macOS and Linux). - Lightweight: a shell hook and symbolic links, with no app, daemon or proxy. A plain `claude` or `codex` launch reads a cached script in the shell and skips clausona; on an M4 Pro Mac, `claude --version` took a median 25 ms with the hook and 25 ms without it. clausona's own process (about 0.3 s) runs on the first launch after a change such as `csn use`, on every launch of an API profile, and on every `clausona run`, routed or not. - A running session never moves to another account. `clausona use` and `clausona run <profile>` run on the account the user names, and a route picks one only when a run starts. Keep the accounts to what they're for, such as work and personal use, within Anthropic's and OpenAI's terms. - Dynamic routing: `clausona route add main` makes a route of every Claude Code account (`--tool codex` for Codex). `clausona run --route main` starts on one of them that still has plan quota left, and says on stderr which one. After `clausona route dir <dir> <route>` or `clausona route default <route>`, plain `clausona run claude` or `clausona run codex` takes a route too. See [dynamic routing in the Claude Code guide](https://larcane97.github.io/clausona/guides/multiple-claude-code-accounts/#routing) and [Routing](https://github.com/larcane97/clausona/blob/main/docs/routing.md). - Extensions: `clausona skills`, `clausona mcp`, `clausona hooks` and the dashboard's Extensions screen list the skills, MCP servers and hooks that each Claude Code and Codex account loads in a project. An item can be turned off or deleted, after a confirm step and behind a backup, and `undo` puts the last change back. See [the Extensions guide](https://larcane97.github.io/clausona/guides/claude-code-skills-mcp-hooks/) and [Extensions](https://github.com/larcane97/clausona/blob/main/docs/extensions.md). ## Guides - [How to use multiple Claude Code accounts on one machine](https://larcane97.github.io/clausona/guides/multiple-claude-code-accounts/): CLAUDE_CONFIG_DIR by hand, what must not be shared, and the clausona way, including shared conversation history, and dynamic routing, which picks the account each `clausona run` starts on. - [How to use multiple Codex CLI accounts on one machine](https://larcane97.github.io/clausona/guides/multiple-codex-cli-accounts/): one CODEX_HOME per account, which files to share, and which must stay with each account, and dynamic routing for Codex accounts. - [Run parallel Claude Code agents on different accounts and models](https://larcane97.github.io/clausona/guides/parallel-claude-code-agents/): subagents run out of sight and all spend one account. Instead, run each agent as a separate Claude Code session with `clausona run <profile>` in any runner (a terminal, tmux, herdr, Orca or Superset), on an account or an API model such as DeepSeek. The superset-fleet, herdr-fleet, orca-fleet and tmux-fleet plugins add a coordinator in each runner: one Claude Code or Codex session plans, routes by rules and usage, checks and retires workers. Install only the one for your runner. - [Claude Code와 Codex 계정 여러 개를 동시에 쓰는 법](https://larcane97.github.io/clausona/ko/): the same guide in Korean. - [Run parallel Claude Code and Codex agents in herdr](https://larcane97.github.io/clausona/guides/parallel-agents-in-herdr/): the herdr-fleet plugin. One session in a herdr pane creates a worktree per task with `herdr worktree create`, starts `clausona run <profile>` in each pane, waits with `herdr agent prompt --wait`, checks the pushed branch and, once you approve, removes the worktree without force. - [Run parallel Claude Code and Codex agents in Orca](https://larcane97.github.io/clausona/guides/parallel-agents-in-orca/): the orca-fleet plugin. One session creates an Orca worktree per task, starts `clausona run <profile>` in its terminal (overriding the CODEX_HOME Orca sets), waits with `orca terminal wait --for tui-idle`, checks the pushed branch and, once you approve, removes the worktree, which also deletes its local branch. - [Run parallel Claude Code and Codex agents in tmux](https://larcane97.github.io/clausona/guides/parallel-agents-in-tmux/): the tmux-fleet plugin, for plain tmux 3.2 or newer on macOS, Linux or WSL, with no GUI, so it also works over SSH. One session in a tmux pane adds a git worktree per task on branch `fleet/<task>`, opens a tmux window labelled with the task and profile, and types `clausona run <profile>` into it. Each worker reports its state: a Claude Code worker through hooks passed with `--settings` from a file in its worktree, a Codex worker by ringing the terminal bell, which a tmux hook records. The session waits on those events, checks the pushed branch and, once you approve, removes the worktree with `git worktree remove` without force and closes the window. - [See which skills, MCP servers and hooks Claude Code loads, and turn them off](https://larcane97.github.io/clausona/guides/claude-code-skills-mcp-hooks/): the skills, MCP servers and hooks Claude Code and Codex load in a project, for each account, on the dashboard's Extensions screen or with `clausona skills`, `clausona mcp` and `clausona hooks`. How to turn one off in one project or everywhere, delete it behind a backup, and undo the change. - [Claude Code 에이전트를 여러 계정과 모델로 병렬 실행하기](https://larcane97.github.io/clausona/ko/parallel-agents/): the parallel agents guide in Korean. - [herdr에서 Claude Code와 Codex 에이전트를 계정별로 병렬 실행하기](https://larcane97.github.io/clausona/ko/parallel-agents-in-herdr/): the herdr guide in Korean. - [Orca에서 Claude Code와 Codex 에이전트를 계정별로 병렬 실행하기](https://larcane97.github.io/clausona/ko/parallel-agents-in-orca/): the Orca guide in Korean. - [tmux에서 Claude Code와 Codex 에이전트를 계정별로 병렬 실행하기](https://larcane97.github.io/clausona/ko/parallel-agents-in-tmux/): the tmux guide in Korean. - [Claude Code와 Codex가 불러오는 스킬, MCP 서버, hook 확인하고 끄기](https://larcane97.github.io/clausona/ko/skills-mcp-hooks/): the Extensions guide in Korean. ## Project - [GitHub repository and README](https://github.com/larcane97/clausona) - [How it works](https://github.com/larcane97/clausona/blob/main/docs/how-it-works.md) - [Superset fleet use case](https://github.com/larcane97/clausona/blob/main/docs/usecases/superset-fleet.md) - [herdr fleet use case](https://github.com/larcane97/clausona/blob/main/docs/usecases/herdr-fleet.md) - [Orca fleet use case](https://github.com/larcane97/clausona/blob/main/docs/usecases/orca-fleet.md) - [tmux fleet use case](https://github.com/larcane97/clausona/blob/main/docs/usecases/tmux-fleet.md) - [Plan quota](https://github.com/larcane97/clausona/blob/main/docs/plan-quota.md) - [API profiles](https://github.com/larcane97/clausona/blob/main/docs/api-profiles.md) - [Routing](https://github.com/larcane97/clausona/blob/main/docs/routing.md) - [Extensions](https://github.com/larcane97/clausona/blob/main/docs/extensions.md)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

