# Tao finetune kumo anomaly

> Tao finetune kumo anomaly is an open-source agent skill for analytics and data. Detects anomalies in multivariate time series and fine-tunes a diffusion model on your. It is available on Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose.

- **Kind:** Skill
- **Pricing:** Open source
- **Platforms:** Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
- **Website:** https://github.com/NVIDIA/skills/tree/main/skills/tao-finetune-kumo-anomaly
- **Repository:** https://github.com/nvidia/skills
- **GitHub stars:** 4,037
- **License:** Apache-2.0
- **Skill name:** tao-finetune-kumo-anomaly
- **Collection:** https://github.com/NVIDIA/skills
- **Categories:** analytics and data
- **Listing updated:** 2026-11-18

## About

Kumo-Anomaly gives an agent a diffusion-based workflow for finding anomalies in multivariate time series. The model reconstructs randomly masked segments of a signal and scores each timestep by the reconstruction error, and adaptive thresholding (SCS or MACS) turns those scores into anomaly flags. The skill points the agent to the inference function in the Kumo-TS SDK, which handles multi-GPU dispatch and returns the original table with Anomaly and MAE columns added.

The skill also covers fine-tuning the pretrained checkpoint on your own mostly-normal data, with options for validation splits, masking, learning rate, epochs and GPU count, and it describes the output files such as the best checkpoint and per-epoch metrics. A separate reference covers AutoML hyperparameter search through a TAO skill bank.

The agent needs Python 3.12 or newer and uv, and it works on CPU, though a CUDA GPU is recommended. Input must be a numeric CSV with no missing or infinite values, and the pretrained weights are downloaded from Hugging Face on first run.

## When your agent uses it

> Kumo-Anomaly — diffusion-based anomaly detection and fine-tuning for multivariate time series. Use when the user asks to "fine-tune Kumo-Anomaly", "run Kumo-Anomaly inference", "detect anomalies with diffusion", "time series anomaly detection", "finetune Kumo-Anomaly", "use perform_anomaly_analysis_with_diffusion", "automl Kumo-Anomaly", "hyperparameter search Kumo-Anomaly", "hyperparameter optimization" or mentions "curriculum_medium.yaml", "final_model.pth", "kumo-anomaly", "sdk/anomaly_analysis", or "TSDiffuser_Generic".

## What it can access

From a static check of the files on GitHub; nothing was run.

- [Instructions only, no scripts](https://github.com/NVIDIA/skills/tree/HEAD/skills/tao-finetune-kumo-anomaly)
- [Asks for: Read, Bash](https://github.com/NVIDIA/skills/blob/HEAD/skills/tao-finetune-kumo-anomaly/SKILL.md)
- [License: Apache-2.0](https://github.com/NVIDIA/skills/tree/HEAD/skills/tao-finetune-kumo-anomaly)

## Features

- **Diffusion-based anomaly scoring:** Scores each timestep by reconstruction error from masked-segment diffusion modeling.
- **Adaptive thresholding:** Converts anomaly scores to binary labels using the SCS or MACS strategies.
- **Pretrained inference workflow:** Runs the SDK inference function with auto-downloaded weights and configuration.
- **Fine-tuning on custom data:** Warm-starts from a published checkpoint and trains on your own time series CSV.
- **Optional PDF reporting:** Generates an anomaly report PDF through an optional configuration object or YAML file.
- **AutoML hyperparameter search:** Points to a reference for tuning fine-tuning and inference parameters with AutoML.

## Good for

- Detect anomalies in a sensor time series CSV
- Fine-tune the pretrained model on normal machine data
- Generate a PDF anomaly detection report
- Run hyperparameter search for fine-tuning settings

## Install

### Claude Code

Copy the skill folder to `~/.claude/skills/tao-finetune-kumo-anomaly` to use it in all your projects:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.claude/skills &&
rm -rf ~/.claude/skills/tao-finetune-kumo-anomaly &&
cp -R "$tmp/skills/tao-finetune-kumo-anomaly" ~/.claude/skills/tao-finetune-kumo-anomaly &&
rm -rf "$tmp"
```

For one project only, put the folder in `.claude/skills/tao-finetune-kumo-anomaly` inside it instead.

### Codex

Copy the skill folder to `~/.agents/skills/tao-finetune-kumo-anomaly` to use it in every repository:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.agents/skills &&
rm -rf ~/.agents/skills/tao-finetune-kumo-anomaly &&
cp -R "$tmp/skills/tao-finetune-kumo-anomaly" ~/.agents/skills/tao-finetune-kumo-anomaly &&
rm -rf "$tmp"
```

For one repository only, put the folder in `.agents/skills/tao-finetune-kumo-anomaly` inside it instead.

### Cursor

Copy the skill folder to `~/.cursor/skills/tao-finetune-kumo-anomaly` to use it in every project:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.cursor/skills &&
rm -rf ~/.cursor/skills/tao-finetune-kumo-anomaly &&
cp -R "$tmp/skills/tao-finetune-kumo-anomaly" ~/.cursor/skills/tao-finetune-kumo-anomaly &&
rm -rf "$tmp"
```

For one project only, put the folder in `.cursor/skills/tao-finetune-kumo-anomaly` inside it instead.

### Gemini CLI

Run in a terminal (add `--scope workspace` to install it for the current workspace only):

```sh
gemini skills install https://github.com/NVIDIA/skills.git --path skills/tao-finetune-kumo-anomaly
```

Or copy the folder to `~/.gemini/skills/tao-finetune-kumo-anomaly` yourself.

### VS Code

For GitHub Copilot in VS Code, copy the skill folder to `~/.copilot/skills/tao-finetune-kumo-anomaly`:

```sh
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.copilot/skills &&
rm -rf ~/.copilot/skills/tao-finetune-kumo-anomaly &&
cp -R "$tmp/skills/tao-finetune-kumo-anomaly" ~/.copilot/skills/tao-finetune-kumo-anomaly &&
rm -rf "$tmp"
```

For one project only, put the folder in `.github/skills/tao-finetune-kumo-anomaly` inside it instead.

### Other agents

Get the folder:

```sh
git clone --depth 1 https://github.com/NVIDIA/skills.git
```

Then copy `skills/skills/tao-finetune-kumo-anomaly` into your agent's skills folder as `tao-finetune-kumo-anomaly`. Agents that support SKILL.md read one folder per skill.

## FAQ

### What does the input data need to look like?

Columns must be numeric, with no NaN or infinite values. Timestamps, IDs and labels should be dropped before calling the inference API, and there must be at least as many rows as the window length.

### Do I need a GPU?

No. CPU-only use is supported, though it is slower. A CUDA GPU is recommended, especially for fine-tuning.

### Which clients does it run in?

It follows the shared agent skills format and is listed for Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose.

### Where do the model weights come from?

The pretrained weights are downloaded automatically from the nvidia/Kumo-Anomaly Hugging Face repository on first run, or can be fetched in advance.

### How do I install Tao finetune kumo anomaly in Claude Code?

Copy the skill folder from GitHub to ~/.claude/skills/tao-finetune-kumo-anomaly to use it in all your projects, or to .claude/skills/tao-finetune-kumo-anomaly in one project.

### Does Tao finetune kumo anomaly run scripts?

No. It is instructions only, with no scripts. This comes from a static check of the files on GitHub; nothing was run. Read the files before you install it.

### Which AI agents can use Tao finetune kumo anomaly?

Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose. Each reads skills in the SKILL.md format from its own folder; the install steps on this page list them.

### What are the best alternatives to Tao finetune kumo anomaly?

The closest alternatives to Tao finetune kumo anomaly are Rea analysis optimization, XLSX, Ads, Analytics and Customer research.

## Where this comes from

- Description: from the skill's own SKILL.md.

## Alternatives

- [Rea analysis optimization](https://ai-powered.com/skills/rea-analysis-optimization): Rea analysis optimization is an open-source agent skill for developer tools.
- [XLSX](https://ai-powered.com/skills/xlsx): XLSX is a free agent skill for files and documents.
- [Ads](https://ai-powered.com/skills/ads-coreyhaines31): Ads is an open-source agent skill for marketing and seo.
- [Analytics](https://ai-powered.com/skills/analytics-coreyhaines31): Analytics is an open-source agent skill for analytics and data.
- [Customer research](https://ai-powered.com/skills/customer-research): Customer research is an open-source agent skill for marketing and seo.
- [Conversion tracking](https://ai-powered.com/skills/conversion-tracking): Conversion tracking is an open-source agent skill for analytics and data.
- [Onboarding](https://ai-powered.com/skills/onboarding-coreyhaines31): Onboarding is an open-source agent skill for marketing and seo.
- [Attribution](https://ai-powered.com/skills/attribution): Attribution is an open-source agent skill for marketing and seo.

---
Source: https://ai-powered.com/skills/tao-finetune-kumo-anomaly
