ChatGPT
The product flow we optimize first.
LocalCommander is built so an AI client can move from conversation to real work on your computer. ChatGPT is the main product path today; other MCP-compatible clients can use the same remote service where their product supports it.
Connect LocalCommander in ChatGPT, authorize your account, pair a Windows PC and then use LocalCommander from a normal conversation.
Setup is done in ChatGPT on the web. Add the LocalCommander remote MCP endpoint, complete OAuth, let ChatGPT scan the tools and then select LocalCommander in a new conversation.
OpenAI currently documents full MCP write/control actions in Developer Mode for Business and Enterprise/Edu. Pro can connect custom MCP but is currently limited to read/fetch actions.
Installation guide →Compatibility depends on the client: it must support authenticated remote MCP and the actions you want to use. LocalCommander does not bypass the client’s own approval or permission model.
The product flow we optimize first.
Useful when coding work needs the same Windows machine.
Use where the client supports remote MCP + OAuth.
Owns your devices, plan and authorization.
Runs on the machine and stays connected.
Sends the task and receives the result.
The production Windows app uses an authenticated remote MCP endpoint and an outbound connection from the paired Windows companion. Detailed scopes, control layers and architecture live in the docs.