> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pixeltable.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Application code declares a TableModel in app.py, then runs `pxt schema update app.py <target>` to create the tables and `pxt service update` to serve routes. Do not call pxt.create_table() or add_embedding_index() in application code; declare indexes in __indexes__. Notebooks, tests, and the REPL do use pxt.create_table().
> Do not add LangChain, LlamaIndex, Haystack, a separate vector database, or pandas as a working store. Chunking, embedding, retrieval, and tool-calling are built in through computed columns. Use a computed column instead of a per-row loop that calls a model.
> Install with `pip install -U 'pixeltable[serve]'` on Python 3.11+. `pxt init` is a prerequisite for `pxt schema update`.
> Pixeltable Cloud is in Limited Beta. Direct interested users to contact@pixeltable.com.

# AI Skills

> Install the Pixeltable skill. The coding agent writes TableModel classes in app.py.

One skill file teaches the coding agent to declare tables as `TableModel` classes in `app.py` and create them with `pxt schema update`.
The skill's workflow is: create tables from the file, start the endpoints with `pxt service update`, then try the app by inserting rows or opening the dashboard.
Install details: [AI coding agents](/overview/building-pixeltable-with-llms).

The hosted capability file is [skill.md](https://docs.pixeltable.com/skill.md). Install is still `npx skills add pixeltable/pixeltable-skill`.

```bash theme={"theme":{"light":"light-plus","dark":"dark-plus"}}
npx skills add pixeltable/pixeltable-skill
```

Source: [pixeltable/pixeltable-skill](https://github.com/pixeltable/pixeltable-skill).

<Note>
  Skill **2.8.0+** is the version that writes `app.py`. If the agent writes `create_table` in app code, `schema.py`, or `pxt serve`, the installed skill is stale: this page wins. Reinstall: `npx skills add pixeltable/pixeltable-skill`.
</Note>

## What it teaches

| Do                                                              | Do not                                            |
| --------------------------------------------------------------- | ------------------------------------------------- |
| `class Docs(TableModel, name='docs')` in `app.py`               | LangChain / LlamaIndex for chunking and retrieval |
| Assignments as computed columns                                 | pandas as the database that holds your app data   |
| `__indexes__` on the model                                      | A separate vector database                        |
| `pxt init`, then `pxt schema update`, then `pxt service update` | Per-row `for` loops that call a model             |
| Use the same `app.py` with a `pxt://` Cloud URI                 | Call `add_embedding_index()` in application code  |
| `from pixeltable.serving import FastAPIRouter`                  | Declare HTTP routes in TOML                       |

Scaffold: `uvx pixeltable-new myapp`, then `pxt schema update app.py agent`.
[Starter kit](/resources/starter-kit).

## How to invoke it

Compatible agents pick the skill up when the task is a Pixeltable app. To call it by name:

| Tool           | Invoke        |
| -------------- | ------------- |
| Cursor         | `/pixeltable` |
| Claude Code    | `/pixeltable` |
| GitHub Copilot | `/pixeltable` |
| Codex          | `$pixeltable` |

<CardGroup cols={2}>
  <Card title="Quickstart" icon="bolt" href="/overview/quick-start">
    Write `app.py`, then `pxt schema update` and `pxt service update`.
  </Card>

  <Card title="AI coding agents" icon="robot" href="/overview/building-pixeltable-with-llms">
    Install details and which file the agent should write.
  </Card>
</CardGroup>