Set up Claude Code once. The whole team inherits it.
CloudCLI runs your Claude Code team setup in the cloud: MCP servers, skills, and CLAUDE.md context configured at the organization level, with an isolated environment per developer. Cursor CLI, Codex, and Gemini CLI run in the same environments.
When Claude Code lives on laptops, the setup can't travel
Every developer ends up with a different CLAUDE.md, different MCP servers, different keys. The setup is the bottleneck, not the agent.
Claude Code reads its setup from the machine it runs on: ~/.claude.json, CLAUDE.md, MCP config, .env. On a team that means five developers with five slightly different setups, onboarding that starts with a half-day README, and agent output shared by pasting files into Slack. CloudCLI moves the setup into a cloud environment, so the environment itself is the shared artifact.
One Claude Code baseline, shared across the team
Configure once at the organization level. Every developer starts from it.
Connect the team's MCP servers and skills once
Configure MCP servers, shared skills, and CLAUDE.md context at the organization level. CloudCLI applies that baseline to every environment's Claude Code config, so there is nothing to copy between machines.
Each developer opens their own environment
Every teammate gets an isolated container built from the baseline: same tools, same context, their own runtime and files. They sign in with their own Claude subscription or API key; CloudCLI forwards it to the agent instead of storing it as team state.
Sessions run in parallel and persist
Developers run Claude Code, Cursor CLI, or Codex side by side on different tasks. Sessions live in the containers, so agents keep working when laptops close.
From org config to a working developer in minutes
The baseline lives at the organization level. Developers inherit it; they never share a runtime.
Onboard a developer the day they join
Invite a teammate and they open an environment with the stack already in place: the right runtimes, your MCP servers connected, CLAUDE.md loaded, Claude Code installed. No README crawl, no works-on-my-machine. Their first agent session starts the same morning.
Keys stay personal. Config stays shared.
Each developer signs in to Claude Code with their own subscription or API key. CloudCLI forwards credentials to the agent instead of turning them into shared team state, and org-level MCP secrets are encrypted at rest. The team shares configuration, never credentials.
More ways to run agents as a team
The same team setup works for every agent CloudCLI runs, not just Claude Code; the agent-agnostic mechanics are covered on the teams page. If you're paying for GitHub Codespaces today, see the GitHub Codespaces comparison. And when the team goes async, background coding agents pick up tasks from Linear or Jira and come back with finished code.
Frequently asked questions
Common questions from engineering leads setting up Claude Code for a team
Create an organization in CloudCLI, connect MCP servers and shared skills at the organization level, and add your CLAUDE.md context. Every developer you invite gets their own cloud environment with that baseline already applied and Claude Code installed. The setup happens once, not once per laptop.
Yes. Every developer gets their own container with dedicated compute and filesystem. They share the base configuration but run independently: no conflicts, no access to another developer's files or running sessions.
Org-level MCP secrets, such as OAuth tokens and client secrets, are encrypted at rest. Each developer signs in with their own Claude subscription or API key, which CloudCLI forwards to the agent rather than storing as shared team state. Developers can add personal keys without exposing them to the rest of the team.
Yes. Each environment has its own SSH subdomain, so VS Code, Cursor, JetBrains, or any SSH-capable IDE connects directly. The web UI covers everything else, from a desktop browser or a phone.
Yes. Different developers can run Claude Code, Cursor CLI, or Codex on different tasks at the same time, each in their own environment. Sessions persist in the cloud, so agents keep working when developers step away.
Yes. The REST API and the n8n node start agent jobs from Linear, Jira, or any pipeline: a ticket moves to In Progress and an agent starts in its own environment. See integrations for the full set.
Per environment, starting at EUR 7/month, with no per-seat minimums. Each developer runs their own environment, and you add or remove environments as the team changes. Tiers are on the pricing page.
Give the whole team one Claude Code setup
Configure the baseline once. Every developer gets an isolated environment, from EUR 7/month.