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

# LLM and Agent Setup

> Give your LLM or coding agent the Code Storage docs as context so it can use the SDK, HTTP API, and Git workflows correctly.

Code Storage is designed to be used from automated systems. When you are asking an LLM or coding
agent to write against the API, give it the docs first so it can look up current SDK methods,
request shapes, authentication requirements, and Git behavior.

There are two supported ways to do this:

* **Use `llms.txt`** when your tool can load a URL as context.
* **Use the Code Storage skill** when your tool supports skill files or local instruction bundles.

## Option 1: Add `llms.txt`

Give your LLM or agent this URL as context:

```text theme={"theme":{"light":"github-light","dark":"min-dark"}}
https://code.storage/docs/llms.txt
```

`llms.txt` is a compact, model-readable version of the documentation. It is the simplest option
when your tool has a field for URLs, external context, or project instructions.

Use it when you want the model to answer questions like:

* Which SDK method creates a repository?
* What parameters does `createCommit` accept?
* How should I authenticate HTTP API requests?
* How do Git remotes, branches, commits, and archives work in Code Storage?

In your prompt, be direct about the context you added:

```text theme={"theme":{"light":"github-light","dark":"min-dark"}}
Use https://code.storage/docs/llms.txt as the source of truth for Code Storage.
Build against the documented SDK and API behavior. Request pages with the `Accept: text/markdown` header.
```

## Option 2: Install the Code Storage skill

Install the Code Storage skill from the SDK repository:

```bash theme={"theme":{"light":"github-light","dark":"min-dark"}}
npx skills add pierrecomputer/sdk
```

This downloads the skill file from the current SDK repo and installs it into your local agent
skills setup. Restart or reload your agent after installation, then ask it to "use the Code Storage
skill" when working with Code Storage.

The skill includes focused instructions for using Code Storage from an agentic workflow. It is a
good fit for tools that keep reusable skills, project knowledge, or local instruction packs across
sessions.

Use the skill when you want the model to:

* Choose between the SDK, HTTP API, and Git protocol for a task.
* Generate implementation code using the right method names and options.
* Follow Code Storage conventions while creating repos, branches, commits, and files.
* Reuse the same Code Storage guidance across multiple prompts or projects.

## Choosing an option

Use `llms.txt` for quick setup and one-off sessions. Use the skill when your tool supports local
skills and you want the Code Storage instructions to travel with the agent.

You can also use both. Add `llms.txt` as live documentation context, then install the skill so the
agent has stable task instructions for how to apply those docs.