Give your agents access
to your machines.
Connect your AI assistant through MCP to work on code, run commands, and check results in your own environment.
Your tools.
Your environment.
Work on real projects
Ask your agent to inspect a repo, edit code, run tests, and check the result on your machine. Your existing files, dependencies, and command-line tools are already there.
Use the assistant you choose
Connect from ChatGPT, Perplexity, Gemini Spark, and other clients that support remote MCP with OAuth. Custom connector availability depends on your app, plan, and workspace settings.
Follow the work
Agents can start processes, read their output, send input, and stop them. Build, debug, and automate from the conversation.
From conversation
to execution.
1. Add your machine
Install RC and enroll the machine you want your agent to use. RC runs in the background so it is available while the machine is online.
2. Connect your agent
Add RC as a custom MCP connection in your assistant and sign in with your passkey. Connect with MCP →
3. Choose its access
Use Observe for read-only machine information, or explicitly approve Terminal access for selected machines so your agent can run commands.
You choose
where agents work.
Approve with your passkey
Terminal access requires your approval and a specific selection of machines. Your Node checks execution authority locally through RC Lock.
Keep access visible
Review your MCP connections and revoke access from RC. Agents use the permissions you grant to work on your machines.
Understand the data path
MCP commands and output pass through hosted RC memory while in use; RC does not persist them. Your connected AI app handles the information it receives under its own data policies.
Step in when you need to
Open a browser terminal or use the RC CLI to work on your machines yourself. Documentation →