Open-source MCP server for temporary remote terminal troubleshooting with your existing AI agent. The remote user runs the client from a one-command invitation, approves access, sees commands and outp
Power your agent
The publisher's own URL. Not a PlugMCP endpoint.
Host a copy with our keys and logs.
https://agenterm.net/mcpThe Claude button above opens Claude with the name and URL filled in. You still confirm it there.
claude mcp add --transport http agenterm https://agenterm.net/mcpRun in a terminal.
{ "mcpServers": { "agenterm": { "command": "npx", "args": [ "-y", "mcp-remote", "https://agenterm.net/mcp" ] } }}claude_desktop_config.json
Check the JSON shape in your browser with the MCP configuration checker. Prefills only this listing's public metadata — it does not test the connection.
Remote URL or a command on your machine? How the two installs differ
Run on PlugMCP
Keep the publisher's tools behind your own key, logs, and install URL.
LaunchFrom the publisher
README is not crawled yet. Open the repository
Open-source MCP server for temporary remote terminal troubleshooting with your existing AI agent. The remote user runs the client from a one-command invitation, approves access, sees commands and output, and can disconnect at any time.
{ "mcpServers": { "agenterm": { "url": "https://agenterm.net/mcp", "headers": {} } }}Overview
Open-source MCP server for temporary remote terminal troubleshooting with your existing AI agent. The remote user runs the client from a one-command invitation, approves access, sees commands and output, and can disconnect at any time.
8 tools
FAQ
| Run a command on a device. Use setup to create a terminal or session_id to reuse one. Omit both for an anonymous terminal, which closes on early completion or stays open if the wait expires. Commands can continue after the wait expires. Use session_read for more output and stop task-owned terminals |
| list_client | List online devices. A positive wait_ms returns early when any device is online. An empty clients array means no device was online when the call returned. An error means the check failed. |
| p2p_transfer_start | Start the remote P2P helper and return after it registers. Run the private bootstrap script for the Agent's execution environment. Use the executable path it prints to check for connected before copying. Repeated calls reuse the device's active connection. |
| session_control | Send input, interrupt, or EOF to an existing session, or stop its terminal. Check client_info.session_controls before interrupt or EOF. Avoid retrying input or signals when delivery is uncertain. After stop, use session_read to check for closed and drain output. |
| session_list | Find retained sessions on online devices. Use status=idle to find reusable terminals or status=live to inspect occupied terminal slots. |
| session_read | Read session output from a byte cursor. Pass command_id to reject reads after a newer command starts. Use status to check completion and drain has_more even after idle or closed. |
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
No required environment variable is listed on this record.
Agenterm is reached over HTTP.
Last checked 2026-11-21. Open-source MCP server for temporary remote terminal troubleshooting with your existing AI agent. The remote user runs the client from a one-command invitation, approves access, sees commands and output, and can disconnect at any time.