Teach your agent to call the solver.
One script, the entire loop: mesh the part locally, upload the geometry, choose a material, run AI filling, read the peak cavity pressure. Every later example is this loop repeated with a decision wrapped around it - get it working once and the rest is arithmetic.
Swap the bundled STL for your own and it is your part, in about fifteen minutes. The result is written viewer-ready, so example 01 opens it without conversion.
Read the platform quickstart at toolkit.simcon.ai and set it up in this folder, following the setup page it links to. Then run one filling simulation on the bundled part and report the peak cavity pressure.
Watch one run go through
This is the actual sequence the example runs, at the speed it happens. Everything after the first login is unattended.
Your agent writes the code. You decide whether the result makes sense.
Scaffold, install, run
The first platform call opens a browser once for the login. After that there is no token to manage.
$ pip install ".[cloud-quickstart]"$ python examples/cloud-quickstart/run_filling.py
Output lands in build/ - the meshed part and the result file.
Two things worth knowing
Meshing and the platform calls belong in the same Session.user_based(). An inner session releases the licence when it exits, and the outer work then fails in a way that looks like an entitlement problem. Pass manage_session=False to helpers that would open their own.
Temperature in K, pressure in Pa, lengths in mm, flow in cm³/s - the same contract on every endpoint. If your own code works in other units, convert once at the edge and range-check there, rather than carrying two conventions around.
Units across the API · temperature K · pressure Pa · spatial mm · flow cm³/s
Both paths are supported. This one is the default.
Stay on cadmould.cloud
It wraps the per-simulation surface - geometry, materials, running a filling simulation, downloading the result - and handles the login. For one-off runs and most automation, nothing else is needed.
Next: put a decision around the loop
One run answers nothing on its own. Example 03 runs many and scores them into a gating decision.