# Airbyte

MCP server for Claude Code, Codex, Cursor and five other agent environments. Wraps PyAirbyte so an agent can pull data from Airbyte connectors without leaving the project.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 342 repo stars
- Author: airbytehq
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/airbytehq/PyAirbyte/tree/HEAD/airbyte/mcp
- Updated: 2026-10-28
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/airbyte

## What it does

The server is hosted over HTTP at `https://mcp.airbyte.com/mcp`, so nothing needs to be built locally.

Under the hood sits PyAirbyte, a Python library for running Airbyte connectors. The agent gets sources and destinations, moves data through a cache, and answers questions about it in the chat. Since version 0.29.0, Python connectors install through `uv` rather than `pip`.

Connectors install three ways: declarative YAML sources download from `connectors.airbyte.com`, Python packages are pulled in automatically, and Docker images are available with `docker_image=True`. There is no orchestration, scheduling, logging or alerting — this is a connector access layer, not a replacement for the full Airbyte platform.

## Requirements
- Needs: Python and uv (commands run via uvx)
- Remote server: `https://mcp.airbyte.com/mcp`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http airbyte https://mcp.airbyte.com/mcp
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add mcp --url https://mcp.airbyte.com/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "mcp": {
            "url": "https://mcp.airbyte.com/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "mcp": {
            "type": "http",
            "url": "https://mcp.airbyte.com/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http mcp https://mcp.airbyte.com/mcp
```

## When to use it
- connect Airbyte connectors to an agent without a separate server
- pull data from a source into a cache and show it in chat
- run a Python connector on a different Python version with `use_python`
- test connectors while developing them
- use local connectors with a CLI on `PATH`

## Example prompts
- Add the `source-github` source and show recent commits from `airbytehq/PyAirbyte`
- Load data from `source-shopify` into a Postgres cache and list the top 10 orders this month
- Check if the local `source-faker` connector runs through `uv`

## Similar
- [signals-scout-data-pipelines](https://affpapa.org/en/nevibecode/skills/signals-scout-data-pipelines.md) — skill · 317 installs
- [recoup-platform-api-access](https://affpapa.org/en/nevibecode/skills/recoup-platform-api-access.md) — skill · 5500 installs
- [suggesting-data-imports](https://affpapa.org/en/nevibecode/skills/suggesting-data-imports.md) — skill · 401 installs
- [dagster-expert](https://affpapa.org/en/nevibecode/skills/dagster-expert.md) — skill · 6500 installs
- [turbo-doctor](https://affpapa.org/en/nevibecode/skills/turbo-doctor.md) — skill · 5100 installs
- [data-researcher](https://affpapa.org/en/nevibecode/agents/voltagent-data-researcher.md) — subagent · ★ 6505 repo stars

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
