Agent Skills
› managedcode/dotnet-skills
› mlnet
mlnet
GitHub指导在.NET应用中集成ML.NET进行机器学习模型训练、评估、推理及部署。涵盖数据准备、特征工程、Model Builder使用及生产环境下的模型版本管理与验证，确保训练与推理分离并符合业务场景。
Trigger Scenarios
在.NET应用中集成机器学习功能
从本地数据训练或重新训练ML.NET模型
审查推理管道、模型加载或AutoML生成的代码
Install
npx skills add managedcode/dotnet-skills --skill mlnet -g -y
SKILL.md
Frontmatter
{
"name": "mlnet",
"description": "Use ML.NET to train, evaluate, or integrate machine-learning models into .NET applications with realistic data preparation, inference, and deployment expectations. USE FOR: ML.NET integration; local model training or retraining; inference pipelines, model loading, evaluation, and deployment review. DO NOT USE FOR: unrelated stacks; generic tasks that do not need this specific guidance. INVOKES: inspect the repository context, edit targeted files, and run relevant build, test, lint, or validation commands when changes are made.",
"compatibility": "Requires ML.NET, Model Builder, or ML.NET CLI scenarios."
}
ML.NET
Trigger On
- integrating machine learning into a .NET application
- training or retraining ML.NET models from local data
- reviewing inference pipelines, model loading, or AutoML-generated code
Workflow
- Start from the prediction task and data quality, not the algorithm or package list.
- Separate training code from inference code so the production path stays lean and predictable.
- Review feature engineering, normalization, label quality, and evaluation metrics before trusting model output.
- Use Model Builder or the ML.NET CLI when they speed up exploration, but inspect the generated C# before treating it as production architecture.
- Plan how the model is loaded, versioned, and refreshed in the application lifecycle.
- Validate with representative datasets and explicit evaluation, not only with a sample that happens to run.
Deliver
- ML.NET pipelines that fit the prediction task
- production-usable inference integration
- evaluation evidence tied to the business scenario
Validate
- model quality is measured, not assumed
- training and inference responsibilities are separated
- deployment and versioning expectations are explicit
References
- patterns.md - Data loading, training pipelines, evaluation metrics, deployment strategies, and feature engineering patterns
- examples.md - Complete examples for sentiment analysis, price prediction, image classification, anomaly detection, recommendations, clustering, fraud detection, text classification, object detection, and AutoML
Version History
- 7ab7f03 Current 2026-07-25 05:22
