Skip to content

Current limitations

  • Commands and started programs run non-interactively (no stdin). Drive GUI apps with the screenshot and UI automation tools instead.
  • A program that detaches itself from its process tree (setsid, daemonising, or on Windows CREATE_BREAKAWAY_FROM_JOB) is beyond the reach of stop_process and run’s timeout.
  • Started programs belong to the agent: any paired bridge can see and stop them, and they stop when the agent stops.
  • Output is decoded as UTF-8. Non-ASCII output from Windows tools using a legacy code page may appear garbled.
  • read_file is for text. Binary content is returned with invalid bytes replaced.
  • write_file takes text only. Binary files, and anything from your machine, go through jump push, which you run yourself (or approve when your AI client runs it), so the MCP server never reads local files on an AI’s behalf.
  • A push larger than 4 MiB is written in pieces, so the file is incomplete on the node until the last piece lands. Write to a temporary name and rename it with run if a half-written file would be picked up.
  • Commands run with the permissions of the user running jump-agent.
  • The tray shows pairing codes in a dialog (a notification on Linux).
  • On Windows, an agent started at login or by double-click has no console, so its log is not kept. Start it from a terminal to see the log. Log files come with signed installers (Phase 2).