> ## Documentation Index
> Fetch the complete documentation index at: https://daily-aleix-evals-docs.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# AI-Assisted Development

> Use AI coding tools with Pipecat documentation and source code.

Pipecat offers several ways to set up AI coding tools and feed them accurate, up-to-date context, helping you build voice and multimodal applications faster.

## Make your project agent-ready (`pipecat init`)

If you're building with an AI coding assistant, run `pipecat init` in your project first:

```bash theme={null}
uv tool install "pipecat-ai[cli]"
pipecat init
```

This writes a Pipecat coding-agent guide (`AGENTS.md` + `CLAUDE.md`) and developer guidance (`GETTING_STARTED.md`) into the project. Your agent picks up `AGENTS.md` automatically — it follows Pipecat conventions and scaffolds the app for you with `pipecat create`. Pair it with the Pipecat Context Hub below so the agent works from live API context instead of stale training data.

<Card title="Build Your Next Bot" icon="robot" href="/pipecat/get-started/build-your-next-bot">
  The full agent-driven workflow, start to finish
</Card>

## Pipecat Context Hub (MCP Server)

[Pipecat Context Hub](https://github.com/pipecat-ai/pipecat-context-hub) is a local-first MCP server that indexes Pipecat documentation, code examples, and framework API source on your machine. It runs a local vector database and is optimized for coding workflows. It returns relevant source code and snippets rather than synthesized answers, giving your AI coding tool the raw context it needs to solve implementation problems.

Because it runs locally, you can also add your own repositories (public or private) to the index for a more customized context.

Available tools include `search_docs`, `search_examples`, `search_api`, `get_code_snippet`, and `check_deprecation`.

### Setup

The hub is published to PyPI as `pipecat-ai-context-hub` and runs with [`uv`](https://docs.astral.sh/uv/) — no clone required. Build the local index once:

```bash theme={null}
uvx pipecat-ai-context-hub refresh   # build local index (~2 min first run)
```

### Add to your coding tool

<Tabs>
  <Tab title="Claude Code">
    ```bash theme={null}
    claude mcp add pipecat-context-hub -- uvx pipecat-ai-context-hub serve
    ```
  </Tab>

  <Tab title="Codex">
    ```bash theme={null}
    codex mcp add pipecat-context-hub -- uvx pipecat-ai-context-hub serve
    ```
  </Tab>

  <Tab title="Cursor">
    Add to your Cursor MCP config:

    ```json theme={null}
    {
      "mcpServers": {
        "pipecat-context-hub": {
          "command": "uvx",
          "args": ["pipecat-ai-context-hub", "serve"]
        }
      }
    }
    ```
  </Tab>

  <Tab title="VS Code">
    Add to `.vscode/mcp.json`:

    ```json theme={null}
    {
      "servers": {
        "pipecat-context-hub": {
          "command": "uvx",
          "args": ["pipecat-ai-context-hub", "serve"]
        }
      }
    }
    ```
  </Tab>
</Tabs>

A newly added MCP server loads when your coding tool starts a session, so add it before opening the session you want to use it in.

### Use it without MCP

The same tools are available as one-shot CLI commands, so you can query the index from a shell with no MCP setup — handy in a session where the MCP server isn't connected yet:

```bash theme={null}
uvx pipecat-ai-context-hub check-deprecation PipelineTask
uvx pipecat-ai-context-hub search-api "WebsocketServerParams"
uvx pipecat-ai-context-hub status            # index health / freshness
```

Each command prints the tool's JSON to stdout. The package name is `pipecat-ai-context-hub`; the command and MCP server are named `pipecat-context-hub` (both spellings of the command resolve once installed).

See the [repo docs](https://github.com/pipecat-ai/pipecat-context-hub) for additional configuration options, including pinning to a specific Pipecat version.

## CLAUDE.md in the framework repo

Separate from the project guide `pipecat init` writes, the [Pipecat framework repository](https://github.com/pipecat-ai/pipecat) ships its own [`CLAUDE.md`](https://github.com/pipecat-ai/pipecat/blob/main/CLAUDE.md) with architecture guidance, coding conventions, and development patterns for the framework codebase itself. When you work in a cloned copy of the Pipecat repo — for example, contributing to the framework — Claude Code reads it automatically.

## Markdown docs

Every page on this docs site has a **Copy page** button (in the top-right area) that copies the page content as markdown. This is useful for pasting into any AI chat or coding tool to provide context about a specific topic.

## llms.txt

Pipecat docs support the [llms.txt standard](https://llmstxt.org/), which provides a machine-readable index of documentation content. This is useful for tools that can ingest documentation in bulk.

* [`https://docs.pipecat.ai/llms.txt`](https://docs.pipecat.ai/llms.txt) — structured index of all pages
* [`https://docs.pipecat.ai/llms-full.txt`](https://docs.pipecat.ai/llms-full.txt) — full documentation content in a single file