Choose the agent. Keep the project.
Run supported AI coding agents against the same repo, shell, files, Git state, and MCP setup. No fresh environment for every tool.
Stop rebuilding setup for each agent
CloudCLI keeps the project surface stable while the agent changes: repo, shell, files, Git state, MCP, and provider access stay put.
Supported agents, one setup
Claude Code, Cursor CLI, Codex, Gemini CLI, and OpenCode run around the same project surface.
Bring the subscriptions you already pay for
Connect your provider access. CloudCLI adds no model markup.
Repo, shell, Git state, and MCP stay put
The project context stays in the environment while the agent changes.
No duplicate setup per tool
Start, resume, and inspect agent work without rebuilding context for each CLI.
Keep repo, shell, Git state, and MCP in one place
Start the next run in the agent that fits. The project context stays attached to the environment.
Use the right agent without moving the work
Claude Code, Cursor CLI, Codex, Gemini CLI, and OpenCode operate around the same project surface. Start the next run with the agent that fits; the files, shell, Git state, and MCP servers stay put.
Bring the subscriptions you already pay for
CloudCLI is not a model reseller. Connect the provider access each agent needs and run it inside the environment. There is no model markup, and keys are not turned into shared team state.
Try new agents without migrating the team
Agents change quickly. CloudCLI keeps the environment stable so your team can test better tools without moving repos, MCP setup, or active work.
Keep the environment stable
Standardize the project surface without standardizing on a single agent vendor.
No model markup
You pay CloudCLI for the cloud environment, not a CloudCLI-branded model.
One place to get back to work
One repo, one MCP configuration, one way back into active sessions.
Before you hand work to an agent
An AI coding agent is a tool like Claude Code, Cursor CLI, Codex, or Gemini CLI that reads your repo, plans changes, edits files, and runs commands, rather than only suggesting completions in an editor. Agents work best in an environment that holds the repo, shell, credentials, and MCP servers. That environment is what CloudCLI provides.
CloudCLI supports Claude Code, Cursor CLI, Codex, Gemini CLI, and OpenCode through its provider layer.
Yes. Claude Code, Cursor CLI, Codex, Gemini CLI, and OpenCode are all installed by the current hosted environment image, ready to run with your own keys.
Yes. CloudCLI is built around bring-your-own-provider access. You connect the credentials or account flow required by the agent you want to run.
No. CloudCLI does not add model markup. You pay CloudCLI for the cloud development environment, not for a CloudCLI-branded model.
Yes. The environment keeps the repo, shell, files, Git state, MCP configuration, and project context in place while you use the supported agent providers around it.
It depends on the task, and the honest answer changes month to month. That is the argument for an agent-agnostic environment: keep the repo, shell, files, and MCP configuration stable in CloudCLI, then choose the supported agent that fits the next run with your own keys and no migration.
Start with one repo. Add agents as you need them.
Keep the project in one persistent cloud environment, then choose the supported coding agent that fits the next job.