Can teammates collaborate in Claude Code or Codex in real time?

Not natively. Claude Code is a single-user tool: one terminal, one session, one account. As of October 2026 there is no built-in way for two people to work inside the same Claude Code session at the same time. The request is open on its issue tracker as #60082, with 24 upvotes, 12 comments and no reply from Anthropic as of October 4, 2026, and earlier requests along the same lines were closed (#27702, #40981, #48828). What you can do falls into four groups: share the setup, share a session afterwards, run more agents, or share the live session itself. The tools that do the last one are compared separately. What Claude Code costs a team, plan by plan, is in Claude Code for teams.

Key takeaways

• Claude Code has no multi-user sessions, and the request for them (#60082) is still open as of September 2026.
• Session sharing links and transcript tools let a teammate read a session afterwards, not steer it.
• Agent Teams and cross-session messaging connect one person's sessions, not several people.
• To let several people steer one agent, teams use Claude Tag in Slack or a shared room such as Poly.

How do the options compare?

Is there a collaborative mode in Claude Code?

No. As of October 2026 Claude Code has no collaborative or multiplayer mode. A session belongs to one account, and that person types every prompt and answers every permission prompt. Anthropic's nearest features each cover part of the job: session links let others read a session, Agent Teams runs several agents for one person, and Claude Tag puts one Claude in a Slack channel. A multiplayer Claude Code session, where several people steer and approve one agent as it works, needs the session to run somewhere everyone can reach. The sections below compare the ways teams get there. Codex's collaboration mode, despite its name, is a setting for one user; it is covered at the end of this page.

How do teams share one Claude Code setup?

The supported way to "collaborate" in Claude Code is to make every teammate's separate session start from the same place:

• Commit a CLAUDE.md so every session inherits the project's context and rules.
• Check in custom slash commands and MCP server configuration.
• Keep a shared prompt library for the tasks the team repeats.

What you get: consistency. Everyone's agent knows the same things. What you don't: anything live. Two people are still in two terminals, and neither can see, approve or stop what the other's agent is doing.

Can Claude Code sessions be shared?

Yes, but only to be read. A shared session is a record of what happened, not a place to steer:

• Claude Code on the web can share a cloud session by link (docs). On Team and Enterprise plans the link can be private or open to the organization; on Pro and Max, private or public to any logged-in Claude user. Viewers see the latest state when they open it, not live updates, and cannot prompt it.
• Lore exports a session, with its prompts, tool calls and diffs, to a private, workspace or public link, and a teammate can fork it into a new session of their own. Free; Team $20 per seat.
• Session Share (a Claude Code skill) exports the session — history, tool calls, file changes — to a redacted JSON file you can send over Slack or email.

What you get: a good hand-off record and a way to review what happened. What you don't: a way to steer. The decisions were made before the link existed; nobody but the original driver approved anything. Every way to share one, including tools that let a teammate type in, is in How to share a Claude Code session.

Can Claude Code agents work together?

Yes, for one person. Claude Code's Agent Teams (experimental; enabled with an environment variable) lets one session act as a team lead that spawns teammate sessions, which split work, share a task list and message each other. Cross-session messaging lets Claude send a message from one of your sessions to another: on the same machine, on your other machines, or in the cloud.

Both are single-player: they connect one person's sessions, and that person owns every approval. They solve parallelism for one person; they do not put a second person in the loop. If your problem is "my cofounder can't see what the agent is doing," more agents make that worse, not better.

What about Claude Tag, Cowork and the Team plan?

Anthropic's own team features share Claude, but not one Claude Code session:

• Claude Tag (June 2026, beta, Slack only, Team and Enterprise plans) puts one Claude in a Slack channel. Anyone can tag it and pick up where the last person left off; it works in a sandbox Anthropic hosts, and a bug fix comes back as a draft pull request. Usage draws on a balance the organization's owner funds. Poly vs Claude Tag compares it with a shared room.
• Claude Cowork applies Claude Code's approach to documents, research and spreadsheets. Its sessions can't be shared; the files they produce can.
• The Team and Enterprise plans include Claude Code with each seat. Each person signs in to their own account and works in their own sessions.

Slack's Slack Code (August 2026) goes further inside Slack: mentioning a supported agent on a task opens a temporary code channel where members prompt it, see diffs and previews, and review its output. Anthropic's Claude is one of the supported agents, alongside Devin, GitHub Copilot and Vercel's.

How can several people steer one Claude Code session?

Run the agent somewhere everyone can reach it, with the session as the shared object. That is what Poly is: a browser room with a git workspace behind it, running Claude Code (and Codex) so a team steers the same agent live — multiplayer agentic coding, in the literal sense.

Concretely:

• One timeline — prompts, replies, tool calls, diffs and receipts stream to every member as they happen.
• Approvals from any seat — in manual mode the agent asks before every edit; the approval card appears on everyone's screen and the first person to decide, decides. The timeline records who.
• Agent questions the room answers together — picks are visible live before anyone submits.
• Attribution and per-person billing — every prompt has an author; every turn's credits are charged to whoever ran it. Nobody needs an API key.
• Revert from any seat — every turn is a commit; any member can undo one from its receipt.
• Sandboxed execution — turns run in isolated containers with an allowlist-only network, so a shared agent is not running on anyone's laptop with anyone's credentials.
• A side channel the agent can't read — threads, group chat and polls for arguing about the plan.
• A view-only watch link — anyone, no account, can watch a room build live.

Honest limits: Poly is a hosted product with its own credits (free to start; watching and approving are always free, running the agent is not). And it is a room, not your terminal — if your workflow depends on local tooling the agent must reach, that has to live in the room's workspace.

Poly is not the only way. HumanLayer lets teammates prompt, interrupt and approve inside one running session for a set time, and OpenClaw 2.0 shares self-hosted sessions between trusted people. Collaborative Claude Code tools, compared covers them all.

Which option should your team use?

• You want everyone's agent to behave the same → share the setup. Do this regardless.
• You need a record of what an agent did for review or hand-off → share the session afterwards.
• You are one person with a big parallelizable task → Agent Teams.
• Your company lives in Slack and has a Claude Team or Enterprise plan → Claude Tag.
• Two or more people need to steer, approve and see the same agent as it works → a shared live session. Nothing else here does this.

Does Codex work the same way?

Everything above applies to OpenAI's Codex CLI as well: it is single-user, its collaboration story is shared configuration (a committed AGENTS.md plays the part of CLAUDE.md), and the same third-party share tools support it.

What Codex's "collaboration mode" is. Codex has a collaboration mode setting, and plan mode is one of its options. It changes how the agent works with you: in plan mode Codex reads the code and proposes a plan before it touches anything. It does not add a second person. The collaboration is between one human and the agent, in one terminal. More in What is Codex collaboration mode?

Poly runs the real Codex CLI in the same rooms as Claude Code. When a room is in plan mode, Codex's own plan arrives as a card on every member's screen, and anyone in the room can approve it, amend it or send it back. Edits and commands wait for an approval from any seat, the same as Claude turns. Each member picks their own model, from Claude, GPT, Grok, Kimi or Muse, and everyone's turns land in one timeline; Grok runs in Codex alongside GPT, and Kimi and Muse run in Claude Code. How a room mixes models has the details.