Appearance
About CulsmaOS
CulsmaOS executes Culsma experiment programs through supported devices or simulators. It manages workstation resources, run state and execution records.
How you can work
Describe your goal in your AI agent. Through MCP, the agent can find experiments, check requirements, prepare configurations, start simulations and read results. You can also complete supported tasks in the desktop interface, or ask the agent to open a page for you to review.
Where two complete methods exist, the task page provides tabs for With your AI agent and In the interface. The manual remembers your choice. Tasks that require a handoff, such as physical loading confirmation, are presented as a continuous sequence.
Four objects to know
| Object | What you use it for |
|---|---|
| Experiment file | Stores a .culs program. Editing the file does not change an already submitted run. |
| Run | One execution of an experiment, with its own identifier, status and results. One file can produce multiple runs. |
| Workstation | The shared environment of devices, layout, positions and resources. The AI agent and desktop should connect to the same workstation. |
| Configuration draft | Proposed device and layout settings. After saving a draft, validate and explicitly apply it. |
Simulation and physical operation
Start with a simulation; no physical instrument is required. Material quantities in simulation results are model records. Completion of a device action does not establish a measured biological result.
The 0.2.0 desktop hardware-run action is disabled. The operator-only loading confirmation boundary remains part of the software contract; the AI agent cannot confirm physical placement on your behalf. See Versions and Supported Features.
Where to start
For a new setup, install and start, connect your AI agent, then complete your first simulation. If your desktop already opens successfully, continue with the connection and experiment guides.