The tuiml command-line interface.

The same operations the Python API and the MCP tools expose, driven from a shell — train, evaluate, benchmark, tune, plot, serve — plus the commands that manage the installation itself.

Commands

  • Modelling: train, predict, evaluate, benchmark,
tune, plot, save, test-statistics.
  • Data: upload, read-data, profile, preprocess,
select-features, generate.
  • Discovery: list, describe, search-source, list-files.
  • Authoring: create-algorithm, read-algorithm,
edit-algorithm, delete-algorithm, get-skeleton.
  • Serving: serve, stop-server, status.
  • Agents: setup, uninstall, mcp, trace.
  • Admin: info, update, restart.

Layout

Every command lives in its own flat module (tuiml.cli.train), exposing exactly one click command whose docstring doubles as its --help text, and is attached to the top-level group by cli.

Connect TuiML to your AI clients, then check what it found:

bash
tuiml setup
tuiml setup --list

Train and serve a model:

bash
tuiml train --algorithm RandomForestClassifier --data iris --target class
tuiml serve model.pkl --port 8000

Modules


benchmark

Benchmark Command - Compare models across datasets via CLI.

0 classes · 1 function
create_algorithm

tuiml create-algorithm: register a user-authored algorithm....

0 classes · 1 function
delete_algorithm

tuiml delete-algorithm: remove a user-authored algorithm from disk....

0 classes · 1 function
describe

Describe Command - Show the parameter schema for any TuiML component.

0 classes · 1 function
edit_algorithm

tuiml edit-algorithm: patch the source of a user-authored algorithm....

0 classes · 1 function
evaluate

Evaluate Command - Evaluate trained models on test data via CLI.

0 classes · 2 functions
generate

Generate Command - Create synthetic datasets for testing and demos.

0 classes · 1 function
get_skeleton

tuiml get-skeleton: print a starter template for a new algorithm....

0 classes · 1 function
info

`tuiml info`: show installation details and check for updates....

0 classes · 1 function
list_cmd

List Command - Browse the TuiML component registry via CLI.

0 classes · 1 function
list_files

List Files Command - Discover built-in and user-authored algorithm files.

0 classes · 1 function
mcp

tuiml mcp: run the TuiML MCP server in the foreground (for debugging)....

0 classes · 1 function
plot

Plot Command - Visualize model behaviour and experiment comparisons.

0 classes · 1 function
predict

Predict Command - Make predictions using trained models via CLI.

0 classes · 2 functions
preprocess

Preprocess Command - Run preprocessing pipelines and atomic stages.

0 classes · 2 functions
profile

Profile Command - Summarise a dataset before training.

0 classes · 1 function
read_algorithm

tuiml read-algorithm: print the source of any registered algorithm....

0 classes · 1 function
read_data

Read Data Command - Preview actual rows from a dataset.

0 classes · 1 function
restart

tuiml restart: kill every running tuiml-mcp; clients respawn it....

0 classes · 1 function
save

Save Command - Copy a trained model to a path of your choosing.

0 classes · 1 function
search_source

Search Source Command - Grep algorithm source files for a pattern.

0 classes · 1 function
select_features

Select Features Command - Rank and keep the most informative features.

0 classes · 1 function
serve

Serve Command - Start a REST API server for model predictions.

0 classes · 1 function
setup

Setup Command - Interactive wizard to connect TuiML to AI agents.

1 class · 22 functions
status

tuiml status: show every running tuiml-mcp instance and which AI client launched it....

0 classes · 1 function
stop_server

Stop Server Command - Shut down model serving API servers.

0 classes · 1 function
test_statistics

Test Statistics Command - Significance tests over experiment results.

0 classes · 1 function
trace

tuiml trace: view or live-tail MCP tool-call traces....

0 classes · 1 function
train

Train Command - Build and train machine learning models via CLI.

0 classes · 2 functions
tune

Tune Command - Search for the best hyperparameters of an algorithm.

0 classes · 1 function
uninstall

Uninstall Command, remove TuiML's MCP wiring from every detected AI client....

0 classes · 9 functions
update

`tuiml update`: upgrade tuiml to the latest PyPI version....

0 classes · 1 function
upload

Upload Command - Register a dataset for use by other TuiML commands.

0 classes · 1 function