Your team's agent setup, already inside the project
A shared cloud dev environment for teams using AI coding agents. Set up MCP servers and skills once; every developer inherits them, isolated.
Share the setup, isolate the runtime
The shared layer is org-wide MCP servers and shared skills. The runtime stays separate: every developer keeps their own container.
Org-wide MCP servers
The tools every project needs, shared once. Personal MCP servers stay personal.
Shared skills
Stored once, then synced into the environments in scope.
Reusable sandboxed environments
New teammates inherit the setup when they start work.
Per-developer containers
Shared setup does not mean a shared runtime.
Configure once. Every developer inherits it.
The organization owns the MCP servers and shared skills. Each developer's container stays theirs.
Configure once at the organization level
Add org-wide MCP servers from Settings, choose stdio or http, and decide whether OAuth uses one organization connection or each user's own token. CloudCLI syncs the server into the environments in scope and records sync results per environment.
New teammate opens a reusable sandbox already configured
When a teammate's environment is created or started, CloudCLI syncs the org MCP servers and shared skills that apply to them. No one copies a private MCP config over chat, and the developer's container stays theirs.
Team baseline. Personal runtime.
CloudCLI separates organization setup from personal setup, so the team standardizes what should be common while each developer works in a separate runtime.
Org-wide baseline
Org MCP servers and shared skills define the baseline every developer starts from.
Personal scope stays personal
User-scoped MCP, OAuth tokens, credentials, and terminal state stay attached to the developer.
No shared runtime
Every developer starts from the same baseline inside their own isolated cloud environment.
FAQ
It is a hosted development setup where the team standardizes the parts that should be common while each developer works in a separate runtime. In CloudCLI, the shared layer is org-wide MCP servers and shared skills, and each developer still opens their own reusable, sandboxed cloud environment.
Standardize the setup, not the runtime. CloudCLI syncs organization-level MCP servers and shared skills into each developer's own environment, so the team starts from the same baseline without sharing one container.
Teams create org-wide MCP servers in CloudCLI settings. Org-wide servers are visible to the organization and sync to environments in scope. User-scoped MCP servers sync only to the environments owned by the developer who created them.
Yes. Team setup can be shared, but each developer still gets a separate cloud environment. Their container, credentials, running processes, and terminal state are not a shared team runtime.
A new teammate inherits the org-wide MCP servers and shared skills that apply to their environment. They still get their own reusable sandbox, container state, credentials, and running processes.
Yes. User-scoped MCP servers and skills sync only to the creator's environments. For OAuth MCP servers, CloudCLI supports either one organization connection or per-user OAuth, so personal tokens do not have to become team-wide state.
Yes, and the same applies to Cursor CLI, Codex, and Gemini CLI. Org-wide MCP servers and shared skills are configured once and sync into every developer's environment, so a Claude Code team stops re-doing setup per laptop. Each developer still works in their own isolated container, and personal credentials stay personal.
Give the team one setup without one shared machine
Share MCP servers and skills once. Every developer inherits them inside their own reusable, sandboxed cloud environment.