Skip to main content

Modern Python Scripting

See More

Use CPython scripting when you want fast automation, data extraction, computational geometry, reporting, debugger support, or package-heavy workflows inside the host.

This page covers the modern CPython path: PythonNet in-process with CAD/BIM hosts. For IronPython, C#/F# scripts, and assemblies, see Execution Overview.

  • CPython
  • Pixi
  • conda-forge
  • PyPI
  • uv
  • pyRevit

How a script runs​

Python runtime by host​

HostPythonPackages
Revit, AutoCAD, Civil 3DCPython 3.14, installed by RevitDevToolPixi: conda-forge and PyPI
Plant 3DThe CPython built into the host, so the version follows the host (Plant 3D 2027 uses 3.13)uv: PyPI only
Other AutoCAD products

Architecture, MEP, Electrical, Mechanical and Map 3D follow the AutoCAD path. See Host settings for the host_name values.

Revit, AutoCAD, Civil 3D​

CPython 3.14 runs in the host process. The environment lives in %APPDATA%\RevitDevTool\pixi-env, and PEP 723 packages install through Pixi.

Plant 3D​

Plant 3D already embeds CPython, so RevitDevTool uses that interpreter instead of starting a second one. Packages install with uv from PyPI into %APPDATA%\RevitDevTool\uv-env\{major.minor}, matching the host's Python version.

Match the host's Python

On Plant 3D, choose packages that publish wheels for the host's CPython version and ABI. A package that only supports a newer Python, or is available only on conda-forge, will not install.

pyRevit CPython engine​

If Pixi cannot run, RevitDevTool falls back to the CPython engine installed by pyRevit. The script and its packages are resolved from that engine, so the Python version is whatever pyRevit provides.

When Pixi or uv cannot run

This is usually company policy blocking the executable, not a problem with your script. If your company allows pyRevit, RevitDevTool falls back to its CPython engine. Otherwise ask IT to allow Pixi. Plant 3D has no fallback and needs uv. See Known Limitations.

pyRevit also controls its own IronPython engine for *_ipy_script.py scripts, which is a separate path described in IronPython — pyRevit Engine.

Entry File​

CPython script folders are discovered through files ending with:

*script.py

A complete script looks like this:

walls_script.py
# /// script
# dependencies = ["polars"]
# ///

import polars as pl
from Autodesk.Revit import DB

doc = __revit__.ActiveUIDocument.Document
walls = DB.FilteredElementCollector(doc).OfClass(DB.Wall)

print(pl.DataFrame({"name": [w.Name for w in walls]}))

More complete scripts, such as data_analysis_script.py, are in PythonDemo/commands on GitHub.

Dependency Metadata​

Dependencies are declared with PEP 723 metadata in the script file:

# /// script
# dependencies = ["polars", "numpy"]
# ///

The runtime parses metadata before execution and prepares the environment with Pixi on every host except Plant 3D, which uses the uv sidecar.

Host Context​

Revit​

Scripts receive __revit__ (UIApplication) and a shared state namespace on sys.__devtool__ for scope-local RevitDevTool state.

AutoCAD family​

There is no __revit__. Use the host Application API instead. See AutoCAD Features for the AutoCAD host context pattern.

For modeless UI or background work in any host, route host API calls through the supported host execution path.

Debugging​

Attach your IDE (VS Code or PyCharm) to the debug port, then run the script from RevitDevTool.

For the full workflow, port settings, .NET attach for scripts/assemblies, and troubleshooting, see Attach Debugger.

IronPython vs CPython Conventions​

These suffixes serve different purposes — do not mix them:

PurposeEntry suffixNotes
Run IronPython automation*_ipy_script.pyAttach your IDE to the IronPython port — Attach Debugger
Test IronPython in-hosttest_*_ipy.pypytest routing only; same Revit engine preference — unittest

*_ipy_script.py is for commands you run from RevitDevTool. test_*_ipy.py is how the pytest plugin routes IronPython unittest files to the host — it is not an execution entry name.

Examples​

All samples live in PythonDemo/commands. Good places to start:

ScriptShows
data_analysis_script.pyCollect model data and analyze it with Polars
debugpy_script.pyDebug a script from your IDE
selectionfilter_script.pySelection filters
visualization_curve_script.pyDraw curves in the active view
logging_format_script.pyLog levels, JSON and stack traces
modeless_script.pyModeless windows
dashboard_ipy_script.pyAn IronPython dashboard with WebView2

Caveats​

  • Package installation can be blocked by local policy, a missing pixi executable (or uv on Plant 3D, which has no fallback), or network access.
  • On Plant 3D, host-owned attach cannot load conda Library\bin activation or a second in-process Python layout — prefer PyPI wheels with private native deps.
  • Host API calls still require valid host context.
  • PythonNet behavior differs from IronPython, especially around .NET interop.