Command line¶
On Windows, use .venv\Scripts\python in place of .venv/bin/python.
wholeboard¶
Starts the server and opens the browser. Also ./run.sh, run.bat or .venv/bin/python -m wholeboard.
| Option | |
|---|---|
--host ADDRESS |
Address to listen on (default 127.0.0.1; 0.0.0.0 for other computers) |
--port N |
Port (default 8770, or port in the settings) |
--no-browser |
Do not open a browser window |
--set-password |
Ask for a password that other computers on the network must enter |
--clear-password |
Remove the network password |
python -m wholeboard.doctor¶
Lists the tools WholeBoard uses, where they were found and their versions.
| Option | |
|---|---|
--install [WHAT ...] |
Install missing tools: tracemaker, freecad, ngspice, libs, platformio, kikit. With no names: all but platformio and kikit |
--force |
Reinstall even if found |
python -m wholeboard.mcp¶
The MCP server, over stdio. No options. See MCP tools.
python -m wholeboard.bench¶
The benchmark. See Run the benchmark.
| Option | |
|---|---|
--only ID,ID |
Only these prompts from bench/prompts.json |
--worker PROVIDER:MODEL |
The worker model, e.g. anthropic:claude-sonnet-5-5 or openai:qwen3 |
--base-url URL |
The OpenAI-compatible server for an openai worker |
--expert MODEL |
The expert model (Claude), or none to turn the expert off |
--timeout MINUTES |
Per prompt (default 45) |
--label NAME |
Name for this run in the table |
--table |
Only rewrite bench/RESULTS.md |
Environment¶
| Variable | |
|---|---|
ANTHROPIC_API_KEY, OPENAI_API_KEY, XAI_API_KEY |
Used when no key is set in Settings |
WHOLEBOARD_HOME |
Where settings and caches live (default ~/.wholeboard) |