Using HoloViews with an AI assistant#

AI assistants (like Claude, GitHub Copilot, or OpenAI Codex) can write HoloViews code much more reliably when they have access to up-to-date, structured documentation instead of guessing from stale training data. This guide shows you how to give your assistant that context, either by pointing it at HoloViews’ generated llms.txt and Markdown docs, or by installing the HoloViz Agent Skills.

What you’ll accomplish#

By the end of this guide you’ll be able to:

  • Install the HoloViz skills so your assistant follows HoloViews best practices automatically

  • Give your AI assistant a curated index of the most useful HoloViews documentation pages

Prerequisites#

  • A python package installer like pip or uvx if you want to install the skills

  • An AI assistant that can read web content or local files (for example Claude, GitHub Copilot, or OpenAI Codex)

Install the HoloViz skills#

The holoviz-skills repository provides a curated collection of Agent Skills for the HoloViz ecosystem, including HoloViews. Agent Skills are folders of instructions and references that teach an assistant how to work with a library correctly.

uvx holoviz-skills@latest install

pip install holoviz-skills
holoviz-skills install

The install command auto-detects the AI tools on your machine. You can also target a specific tool, for example:

holoviz-skills install --claude-code
holoviz-skills install --copilot

You may need to restart your assistant so it picks up the newly installed skills.

Point your assistant at the docs#

HoloViews publishes a machine-friendly index of its documentation that your assistant can fetch directly.

  1. Tell your assistant to fetch the index:

    https://holoviews.org/llms.txt

  2. The index lists the most relevant pages and links to their Markdown versions under:

    https://holoviews.org/markdown/, for example https://holoviews.org/markdown/getting_started/Introduction.md

  3. Ask your assistant to read the pages relevant to your task before writing code. For example, to create a plot you might have it read the Getting Started tutorial and the API reference page for the method you need.

The Markdown pages are intentionally free of the navigation chrome, CSS, and JavaScript of the full HTML site, so they consume far fewer tokens and are easier for an assistant to parse.

Which approach should I use?#

  • Point your assistant at the docs is a good fit for a one-off task or when you can’t install packages, and it always reflects the latest published docs.

  • Install the HoloViz skills gives your assistant packaged, opinionated guidance that works offline, but you need to update it as HoloViews evolves. Once the skills are installed, pointing to the Markdown docs should no longer be needed if you want your agent to be able to fetch the latest docs, as the skills already teach the agent how to do that.