# Looker MCP

> This skill provides instructions for interacting with Looker, BigQuery, and Dataplex via MCP in this environment.

- Skill: `izzyfresh/looker-mcp` (Agent Skill)
- Install (CLI): `npx skillmds@latest add izzyfresh/looker-mcp`
- Raw SKILL.md: https://api.skillmd.com/api/skills/izzyfresh/looker-mcp/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: IzzyFresh (https://skillmd.com/u/izzyfresh)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/izzyfresh/looker-mcp

---


# Looker, BigQuery, and Dataplex MCP Setup

This workspace uses a combination of the official Looker Toolbox and custom Python scripts to provide MCP capabilities for Looker, BigQuery, and Dataplex.

## Configuration Files

The primary configuration files mapping these servers are:
- `~/.gemini/jetski/mcp/config.json` (Primary for Antigravity)
- `~/.gemini/settings.json` (Primary for Gemini CLI)
- `/usr/local/google/home/israelcastillo/gcp-projects/local_mcp_config.json` (Consolidated reference)

## 1. Looker

Looker is configured to use the official **Looker Toolbox** binary for full feature support. Additionally, a `looker-dev` prebuilt is enabled in the `gcp_toolbox` for programmatically managing LookML.

### Standard Looker Tools
- **Command:** `/usr/local/google/home/israelcastillo/gcp-projects/looker-mcp/toolbox_linux`
- **Arguments:** `["--stdio", "--prebuilt", "looker", "--log-level", "error"]`
- **Environment Variables:** `LOOKER_BASE_URL`, `LOOKER_CLIENT_ID`, `LOOKER_CLIENT_SECRET`, `LOOKER_VERIFY_SSL`

This provides a rich set of tools for interacting with dashboards, looks, and the LookML semantic model.

### LookML Management Tools (looker-dev)
The following tools are available when the `looker-dev` prebuilt is enabled:
- **`create_view_from_table`**: Automatically generates a LookML view from a BigQuery table schema.
- **`create_project_file`**: Creates a new LookML file (like a .model or .explore file) in a specified project.
- **`create_project_directory`**: Manages the folder structure within a Looker project.
- **`create_git_branch`**: Prepares the project for changes by creating a new dev branch.

**Usage Example for `create_project_file`:**
```json
{
  "method": "tools/call",
  "params": {
    "name": "create_project_file",
    "arguments": {
      "project_id": "your_project_name",
      "file_path": "models/new_model.model.lkml",
      "content": "connection: \"my_connection\"\\ninclude: \"/views/*.view.lkml\"\\n\\nexplore: new_view {}"
    }
  }
}
```

## 2. BigQuery

BigQuery uses a native Python MCP server for data access.

- **Location:** `/usr/local/google/home/israelcastillo/gcp-projects/looker-mcp/bigquery_mcp_server.py`
- **Execution:** `uv run --with google-cloud-bigquery --with mcp python3 ...`
- **Tools:** `list_datasets`, `list_tables`, `get_table_info`, `execute_sql`

## 3. Dataplex

Dataplex uses a native Python MCP server for catalog search.

- **Location:** `/usr/local/google/home/israelcastillo/gcp-projects/looker-mcp/dataplex_mcp_server.py`
- **Execution:** `uv run --with google-cloud-dataplex --with mcp python3 ...`
- **Tools:** `search_entries`

## Verification

A test script is available at `/usr/local/google/home/israelcastillo/gcp-projects/looker-mcp/test_mcp_servers.py` to verify all three connections. Run it with:
```bash
uv run --with mcp python3 looker-mcp/test_mcp_servers.py
```