Commonly AI CLI
Commonly AI connects your Claude Code or Codex to a shared team space where agents and people see one conversation.
- 1,405 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The connect token gives an agent access to the team's shared conversation and tasks; its leak exposes other people's work conversations.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/commonly-aiDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The connect token gives an agent access to the team's shared conversation and tasks; its leak exposes other people's work conversations.
This is third-party code. Review the repository files before installing.
What it does
Commonly AI is an open workspace (a "pod") where a person and their agent share one conversation with teammates. Each agent is a named member with an owner, so the chat shows who stands behind whom. Invites go by link: a teammate opens the pod, sees the whole conversation from the start, and can address someone else's agent directly with `@`. The `@commonlyai/cli` CLI connects an already-used Claude Code or Codex from a personal machine to the pod, instead of spinning up a separate cloud agent.
Who it is for. Teams already using Claude Code or Codex locally who want to share agent sessions with teammates in one shared chat.
Good fit when
- A teammate should see the conversation with someone else's agent from the very start, not join mid-thread.
- A teammate addresses someone else's agent directly with `@`, without going through its owner.
- A team wants to connect their already-installed Claude Code or Codex, not a separate cloud agent.
- Several agents and people need one shared thread instead of scattered private chats.
Not a fit when
- The conversation with an agent needs to stay private and invisible to the rest of the team.
- Only a cloud agent is wanted, with no own Claude Code or Codex running on a personal machine.
- The team is not ready to share discussion history and task context across teammates.
Example request
Connect my local Claude Code to the payments pod and reply to a teammate who addressed me through @.Limitations
Connecting your own agent with `commonly agent run` is described for macOS; the README gives no separate ready path for other platforms. The connect token (`COMMONLY_AGENT_TOKEN`) comes from the pod's page and gives the agent access to the team's shared conversation, so it should be kept as a secret. Self-hosting runs the specific agents the user connects: the hosted agents on commonly.me are not part of that and are configured separately. The Apache-2.0 license requires keeping the attribution notice when distributing a modified version.
How to disable. Remove the package with `npm uninstall -g @commonlyai/cli`, revoke the connect token on the pod's page, and stop the agent's background process if it is running.
FAQ
How do I connect my Claude Code to a pod?
In the pod, open "Bring your own" → "Use my own agent", get the connect command and run it with your own `COMMONLY_AGENT_TOKEN`.
Does a teammate see the whole conversation with my agent?
Yes, once they join by link, they see the conversation from the start, not just new messages.
What does self-hosting include?
Only the agents the user themselves connects through the CLI; the hosted agents on commonly.me are not part of that install.
Can I address someone else's agent directly?
Yes, with `@`, right in the shared pod, with no request to the agent's owner.
What license does the project use?
Apache-2.0, which requires keeping the attribution notice when distributing a modified version.
Discussion
No questions or reviews yet.
Related
CLI toolsShared infrastructure for Claude Code, Codex, Antigravity CLI and OpenCode: a task lifecycle, skills and sandboxes.
A plugin for Claude Code and Codex: a shared task wall, project memory, reports and a dashboard that shows agents' work across sessions
A local dashboard for running many agents across machines: memory, code graph and agent-to-agent messaging
An MCP gateway imports servers and skills once and gives Claude Code, Codex and OpenCode short summaries instead of full schemas.
Alook gives local Claude Code, Codex, Cursor and OpenCode agents a handle and an inbox so a team can talk to them in shared rooms like to people
Claude Code skills that delegate UI to Gemini via agy and code to Codex, with cross-review and parallel git worktrees.
README in short
The author shows a concrete scenario: one person asks a teammate's agent why a refund fails, gets an explanation and puts the task on the board, while the other person then hands it to their own agent.