Cloud coding agent
Agents and tools · cloud agent, background agent, remote agent
A cloud coding agent is an AI agent that runs a task on a remote server in an isolated environment with a copy of your repo. You can start it and close the laptop: the result arrives as a ready branch or pull request. Codex cloud, Claude Code on the web and Cursor cloud agents work this way.
Cloud agent vs local agent
- Where it runs · a cloud agent runs on the agent vendor’s server; a local one runs on your computer.
- Does it need your computer · a cloud one does not: you can start a task from your phone and close the laptop. A local one works while your computer is on.
- From your phone · you give a cloud agent a task in the phone app or browser, no computer needed. You can also steer a local agent from your phone, through Remote Control in Claude Code and Remote in the ChatGPT app for Codex, but your computer has to be on and online.
- What it sees · a cloud one sees only the repo and what you connect to its environment; a local one sees your files, programs and logins.
- What you get · a cloud one sends a branch or a pull request; a local one edits files in front of you.
- When to use which · cloud for several tasks in parallel and long tasks; local when you need your test stand, browser or quick edits as you go.
Source: Claude Code in the cloud docs
Where it appears · 2
- MCP server and AI agent for 1C · Harness
- Claude Code, Codex, Cursor and Hermes Agent · Harness