Installation
Use Python 3.10–3.13 and a separate environment. This keeps openWFN's dependencies separate from your other projects.
Use a separate environment
python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install --upgrade openwfn
openwfn --version
On Windows PowerShell, activate with .venv\Scripts\Activate.ps1; in Command
Prompt, use .venv\Scripts\activate.bat. With Conda, create an environment
with a supported Python version, activate it, then use its python -m pip.
If Linux reports that ensurepip is unavailable, install your distribution's
matching Python venv package or use a Conda environment.
For development from source, replace the package-install command with:
The normal installation includes file readers, output-property extraction,
local MCP and terminal chat. You do not need separate extras for these features.
Older interop, outputs and mcp install commands still work.
What installation does not supply
A chat model is needed for general conversation, not for file analysis. Use
openwfn chat and /connect for setup guidance. No weights are downloaded and
no model calls are made by installation. See Scientific assistant or
MCP setup.
Gaussian binary .chk conversion needs licensed Gaussian formchk on
PATH. Formatted .fchk files do not need Gaussian. File support and
analysis availability are different; check openwfn FILE capabilities.
Check with packaged examples
The install command also copies the eleven-molecule everyday-QC corpus to
openwfn-examples/everyday-qc/. It does not replace existing files unless
you explicitly permit overwrite.
Reproducing an older analysis
Use the exact version recorded with that work. For the current published release:
Older release pins and their support boundaries are retained in release history, rather than mixed into this setup guide.
Optional trusted-process monitoring is available with the resources extra.
It records runtime/resource evidence; it is not an OS quota or a change to
scientific validation status.