Guide · Agent skill
How to use the Aeon Time Series ML agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The aeon agent skill is a set of instructions and reference files that teaches an AI agent how to do time series machine learning with the aeon Python library. It covers classification, regression, clustering, forecasting, anomaly detection, segmentation and similarity search on time-ordered data such as sensor readings, prices or measurements. This guide explains what the skill does, how to install it, how to trigger it with plain requests, and what to watch for before you rely on it.
Aeon Time Series MLK-Dense-AI/scientific-agent-skills · MIT Download the skill Quick start
- 1Download the skill. Download the aeon skill ZIP from the skill page.
- 2Unzip it into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. Alternatively, upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
- 3Check Python. Confirm you have Python 3.11 to 3.14 available.
- 4Install aeon. Run uv pip install "aeon==1.6.0", adding extras only for the methods you need.
- 5Ask a time series question. Describe your task in plain language, such as classification or forecasting, and mention your time-ordered data.
- 6Review the code. Check the agent's code, data shapes and any downloads before running it on real data.
What the aeon skill makes the agent do
An agent skill is a folder with a SKILL.md file and supporting reference documents that an AI agent loads when a task matches the skill's description. The aeon skill is triggered by time series work: temporal data, sequential patterns, or time-indexed observations that need specialised algorithms beyond standard ML approaches.
Once installed, the agent knows how to pick a suitable aeon algorithm for your task, write code using scikit-learn compatible APIs, and note which methods are stable versus experimental. The skill's reference files list selected classifiers, regressors, clustering methods, forecasters, anomaly detectors, segmenters, distances and transformations.
- Classification and regression, for example with RocketClassifier or RocketRegressor
- Clustering with TimeSeriesKMeans, KShape and related methods
- Forecasting, anomaly detection, segmentation and similarity search (marked experimental)
- Feature extraction with Rocket and Catch22, and distance measures such as DTW
Install the aeon skill
Download the skill ZIP from the skill page and unzip it into one of two locations. For a personal setup, use ~/.claude/skills/. For a single project, use the .claude/skills/ folder inside the project. Alternatively, in the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.
The skill does not install the aeon library for you. Set up the Python library separately, using the pinned version the skill was reviewed against:
- Personal: unzip into ~/.claude/skills/
- Project: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
uv pip install "aeon==1.6.0"Check the requirements before you start
The skill requires Python 3.11 to 3.14 and aeon 1.6.0. You need uv or pip to install packages, and network access for installation and for downloading remote datasets or results. Optional extras depend on the method you choose: stumpy for matrix profiles, and TensorFlow for deep learning estimators.
The skill's examples and library target Python, so it is not suited to non-Python environments.
- Python 3.11 to 3.14
- aeon 1.6.0, installed with uv pip or pip
- Optional: stumpy for matrix-profile examples
- Optional: TensorFlow for deep learning estimators
- Network access for installation and remote datasets
uv pip install "aeon[all_extras]==1.6.0"Trigger the skill with natural requests
You do not need special syntax. Describe the time series task in plain language and mention that the data is ordered over time. The agent will match the request to the skill when it fits the description.
Name the outcome you want, the kind of data, and the language. Stating the goal helps the agent choose between classification, forecasting, anomaly detection and similar tasks.
- "Classify these 500 ECG recordings into normal and abnormal using aeon in Python."
- "Forecast the next 5 days of this daily sales series with an aeon forecaster."
- "Find unusual stretches in this vibration sensor log using aeon anomaly detection."
- "Group these price series into 3 clusters using DTW distance."
- "Where does the behaviour in this sensor trace change? Suggest an aeon segmenter."
Example: a quick classification run
The skill's classification example loads the GunPoint dataset, trains a RocketClassifier and reports test accuracy. The agent adapts this pattern to your own data, so check that your arrays have the shape and labels the estimator expects.
from aeon.classification.convolution_based import RocketClassifier
from aeon.datasets import load_classification
X_train, y_train = load_classification("GunPoint", split="train")
X_test, y_test = load_classification("GunPoint", split="test")
clf = RocketClassifier(n_kernels=10000)
clf.fit(X_train, y_train)
accuracy = clf.score(X_test, y_test)Safety notes: scripts and network access
This skill runs scripts and uses the network. Installation pulls packages, and remote dataset or result downloads need network access. Review what the agent plans to run, especially package installs and any code that fetches data, before approving it in your environment.
The skill's own guidance is also worth knowing. Forecasting, anomaly detection, segmentation, similarity search and visualisation are experimental, and their interfaces may change between minor releases. Some remote archive and TensorFlow training snippets were illustrative and were not executed during the review, so test them on your data first.
- Review install commands and downloads before approving them
- Pin aeon==1.6.0 for reproducibility
- Test remote-data and TensorFlow snippets before relying on them
Tips for better results
State the Python version and whether your series are univariate or multivariate. The skill notes that you should verify estimator capability tags before choosing multivariate, missing-value or unequal-length support. For matrix profiles, install stumpy alongside aeon. For range precision and recall, the skill notes a conflict between the prts package and aeon 1.6, so prefer the AUC metrics or a separately validated environment.
Use stable modules such as classification, regression, clustering, distances and transformations for production work. Treat the experimental modules as prototypes unless you accept interface changes.
- Give the agent your data shape, sampling frequency and goal
- Check an estimator's python_dependencies tag before installing optional packages
- Start with stable modules for production pipelines
What people use it for
- Classifying sensor recordings or ECG traces into labelled categories
- Forecasting short-term values for sales, demand or measurements
- Flagging unusual stretches in machine or server logs
- Grouping price or activity series into clusters with DTW distance
- Locating change points in a sensor trace for segmentation
Frequently asked questions
How do I install the aeon agent skill?
Download the skill ZIP from the skill page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills. Separately, install the library with uv pip install "aeon==1.6.0".
What are the requirements for the aeon skill?
You need Python 3.11 to 3.14 and aeon 1.6.0, plus uv or pip for installation. Stumpy is optional for matrix profiles, TensorFlow is optional for deep learning estimators, and network access is needed for installation and remote datasets.
Is the aeon skill free?
The skill is listed under the MIT licence on this page. The aeon library itself is licensed separately, so check its licence in the project's own documentation before using it commercially.
Does the aeon skill work outside Python?
No. The examples and the library target Python 3.11 to 3.14, so the skill is not suited to non-Python environments.
Which aeon methods are experimental?
Forecasting, anomaly detection, segmentation, similarity search and visualisation are experimental, and their interfaces may change between minor releases. Classification, regression, clustering, distances and transformations are the stable modules recommended for production pipelines.
Should I use aeon or a standard ML tool for my time series?
For simple trend charts or tasks standard ML tools already handle, the skill adds more setup than it saves. Aeon is most useful when you need specialised time series algorithms, such as DTW-based clustering, ROCKET classifiers or matrix-profile anomaly detection, with scikit-learn style code.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the aeon skill if you are a Python data scientist, researcher or engineer who needs specialised time series algorithms and scikit-learn style code. Pick something else if you only need simple trend charts, want non-Python tooling, or need stable interfaces for experimental tasks like forecasting or anomaly detection in production.
Download the skill