SKILLS

Probabl AI Agent Skills

by Probabl

agent-skillsmachine-learning

Overview

Agent skills from Probabl, the scikit-learn maintainers, for rigorous ML pipeline building, evaluation, testing, iteration, auditing, and Python API lookup.

Details

Probabl AI Agent Skills is an official Probabl project published at probabl.ai and in the probabl-ai/skills GitHub repository. The source describes thirteen agent skills from the scikit-learn maintainers for rigorous ML pipeline building, evaluation, testing, iteration, and Python API lookup. The repository catalog identifies probabl-skills version 0.6.0 and lists skills including build-ml-pipeline, evaluate-ml-pipeline, test-ml-pipeline, audit-ml-pipeline, and python-api. The build-ml-pipeline skill declares a pipeline from a data source to a predictor as a skrub DataOps graph, while explicitly scoping out fitting, splitting, tuning, persistence, and evaluation.

When to Use

Use when you want AI-agent guidance for building evaluating testing auditing or iterating on machine-learning pipelines with practices associated with the scikit-learn maintainers. Use when an agent needs structured help with Python API lookup or with ML pipeline steps that should be separated from fitting splitting tuning persistence and evaluation.

Getting Started

Open the official skills page at https://www.probabl.ai/skills/ to review the available Probabl AI Agent Skills.
Inspect the GitHub repository at https://github.com/probabl-ai/skills and its catalog.json to see the published probabl-skills version and listed skills.
Review individual SKILL.md files
such as skills/build-ml-pipeline/SKILL.md
before applying a skill in an agent workflow.
Start with a narrow workflow such as build-ml-pipeline
evaluate-ml-pipeline
test-ml-pipeline
audit-ml-pipeline
or python-api based on the task you want the agent to perform.

Key Features

•Thirteen agent skills described by Probabl for rigorous ML pipeline work and Python API lookup.
•Official GitHub repository under the probabl-ai organization
•with a catalog.json identifying probabl-skills version 0.6.0.
•Includes named skills such as build-ml-pipeline
•evaluate-ml-pipeline
•test-ml-pipeline
•audit-ml-pipeline
•and python-api.
•build-ml-pipeline represents a pipeline from data source to predictor as a skrub DataOps graph and documents exclusions such as fitting
•splitting
•tuning
•persistence
•and evaluation.

Capabilities

•ML pipeline building guidance
•ML pipeline evaluation guidance
•ML pipeline testing and auditing guidance
•Python API lookup support

Last updated Jun 27, 2026