tuiml

<p align="center"> <img src="https://raw.githubusercontent.com/tuiml/tuiml/main/tuiml_logo.png" alt="TuiML Logo" width="320"> </p> <p align="center"><strong>Machine Learning that agents can actually call.</strong></p>

<p align="center"> TuiML is an agent-native ML runtime. Install, connect to your AI agent, and start running real ML workflows — classification, regression, clustering, experiments — all from one structured interface. </p>

<p align="center"> <a href="https://pypi.org/project/tuiml/"><img src="https://img.shields.io/pypi/v/tuiml?style=for-the-badge" alt="PyPI version"></a>  <a href="https://pypi.org/project/tuiml/"><img src="https://img.shields.io/badge/Python-≥3.10-blue?style=for-the-badge&logo=python&logoColor=white" alt="Python versions"></a>  <a href="https://tuiml.ai/docs/getting_started.html"><img src="https://img.shields.io/badge/Docs-tuiml.ai-blue?style=for-the-badge" alt="Documentation"></a>  <a href="LICENSE"><img src="https://img.shields.io/badge/License-BSD--3--Clause-blue.svg?style=for-the-badge" alt="BSD-3-Clause License"></a>  <a href="https://pepy.tech/projects/tuiml"><img src="https://img.shields.io/pepy/dt/tuiml?style=for-the-badge" alt="Downloads"></a> </p>

Why TuiML

Agents can call it — Every algorithm, dataset, and metric ships with a JSON schema. Agents read the schema, call the tool, get structured results. No hallucinated parameters, no wrapper glue.

Agents can discover it — A queryable registry tagged by task, data shape, and benchmarks. Agents browse and pick instead of memorising class names.

Agents can trust it — Deterministic, typed, reproducible outputs. Every call is a loggable, replayable tool invocation you can audit, diff, and trust in production.

Get running in 3 steps

1. Install — one command, installs uv and tuiml globally:

curl -fsSL https://tuiml.ai/install.sh | bash

Already have Python? pip install tuiml works too.

2. Connect your agent — auto-detects Claude Desktop, Cursor, Claude Code, and more:

tuiml setup

3. Ask your agent — in any connected client:

"Train a random forest on my sales data and report the accuracy."

Your agent discovers algorithms, sets parameters from the schema, trains, evaluates, and returns structured results. No glue code.

What's Included

TuiML ships with 13 algorithm families, many originally from Weka, completely rewritten in Python with C++ acceleration for hot paths.

Category	Examples
Trees	RandomForestClassifier, C45TreeClassifier, HoeffdingTreeClassifier, M5ModelTreeRegressor
Bayesian	NaiveBayesClassifier, BayesianNetworkClassifier, GaussianProcessesRegressor
Neighbors	KNearestNeighborsClassifier, KStarClassifier
Linear	LogisticRegression, LinearRegression, SGDClassifier
SVM	SVC, SVR
Neural	MultilayerPerceptronClassifier, VotedPerceptronClassifier
Rules	ZeroRuleClassifier, OneRuleClassifier, RIPPERClassifier, PARTClassifier
Ensemble	BaggingClassifier, AdaBoostClassifier, StackingClassifier, VotingClassifier
Gradient Boosting	XGBoostClassifier, CatBoostClassifier, LightGBMClassifier
Clustering	KMeansClusterer, DBSCANClusterer, AgglomerativeClusterer
Associations	AprioriAssociator, FPGrowthAssociator
Anomaly Detection	IsolationForestDetector, LocalOutlierFactorDetector
Time Series	ARIMA, ExponentialSmoothing, Prophet

Plus preprocessing (scaling, encoding, imputation, SMOTE, text vectorization), feature engineering (selection, extraction, generation), evaluation (metrics, cross-validation, tuning, statistical tests), and 15+ built-in datasets.

MCP Tools

The MCP server exposes 200+ tools agents can call directly. Key workflow tools include tuiml_train, tuiml_predict, tuiml_evaluate, tuiml_experiment, tuiml_tune, tuiml_plot, tuiml_list, tuiml_describe, and tuiml_search. Any component registered with @classifier, @regressor, or @transformer is automatically discoverable through these tools.

For manual setup, add this to your client's MCP config:

{
    "mcpServers": {
        "tuiml": { "command": "tuiml-mcp" }
    }
}

Component Registry

Browse all registered algorithms, transformers, and metrics from the local registry:

from tuiml.hub import registry

classifiers = registry.list("classifier")
regressors = registry.list("regressor")

Building Custom Components

Register your own algorithms and they become instantly available through the Python API, CLI, and MCP server.

from tuiml.base.algorithms import Classifier, classifier

@classifier(tags=["custom"], version="1.0.0")
class MyClassifier(Classifier):
    def __init__(self, k=5):
        super().__init__()
        self.k = k

    def fit(self, X, y):
        self.classes_ = np.unique(y)
        self._is_fitted = True
        return self

    def predict(self, X):
        self._check_is_fitted()
        return predictions

Documentation

Full documentation is available at tuiml.ai/docs, including getting started guides, API reference, and tutorials.

License

BSD 3-Clause License. See LICENSE for details.

Citation

@software{tuiml2026,
    title={TuiML: Machine Learning that agents can actually call},
    author={Verma, Nilesh and Bifet, Albert and Pfahringer, Bernhard},
    year={2026},
    url={https://tuiml.ai}
}

Links

• Website
• Documentation
• API Reference
• GitHub
• PyPI
• Changelog

Star History

<a href="https://www.star-history.com/?repos=tuiml%2Ftuiml&type=date&legend=top-left"> <picture> <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/image?repos=tuiml/tuiml&type=date&theme=dark&legend=top-left" /> <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/image?repos=tuiml/tuiml&type=date&legend=top-left" /> <img alt="Star History Chart" src="https://api.star-history.com/image?repos=tuiml/tuiml&type=date&legend=top-left" /> </picture> </a>