gemini-cli-extensions/looker
claude-codeSkills for Looker
Looker Agent Skills is an open-source plugin by Google that connects AI agents like Claude Code, Antigravity, and Codex to Looker, so you can explore data, manage dashboards, and write LookML using natural language. Think of it as a prompt-driven control layer over your Looker instance.
Our Review
Released by Google under Apache-2.0, Looker Agent Skills reached v0.3.5 in June 2026 and lets any supported AI agent operate a Looker instance through plain-English prompts. The draw is simple: instead of clicking through Explores or hand-writing LookML, you describe the data question or the model change you want.
What Looker Agent Skills adds to your AI agent:
- •Natural-language data exploration ask questions like "show me the top 10 products by sales last month" and the agent queries your Looker models directly.
- •Dashboard and content management create dashboards, find existing ones, and list their tiles by describing the result you want.
- •LookML development generate or update views and measures from a prompt, aimed at data engineers who maintain the semantic layer.
- •Two skill bundles a Looker bundle for data discovery and business intelligence, plus a Looker Development bundle for LookML authoring and administration.
- •Looker API connection authenticates to your instance with LOOKER_BASE_URL, a client ID, and a client secret, with optional flags to surface hidden models, explores, and fields.
Compatibility:
Looker Agent Skills installs into multiple host agents rather than running on its own. It supports Antigravity CLI and Antigravity 2.0, Claude Code v2.1.94 or higher (via the marketplace add and plugin install flow), and Codex v0.117.0 or higher. You can also register the skills through the open npx skills add command. Each host needs the Looker environment variables set before the skills can connect.
Limitations:
The project is in beta (pre-v1.0) and the maintainers warn that breaking changes are possible until the first stable release, so production teams should pin a version and track the releases page. It is not a standalone BI tool: you need an existing Looker instance and valid API credentials, which rules out anyone who is not already a Looker customer. With 47 GitHub stars it is early, and support runs through a feedback form and GitHub issues rather than a paid tier. Each host agent has its own minimum version requirement, so check compatibility before installing.
Our Verdict
Looker Agent Skills is a genuinely useful bridge for teams that already run Looker and want to drive it from an AI agent instead of the UI. As an official Google package under Apache-2.0, it carries more trust than a random community connector, and the natural-language interface to data, dashboards, and LookML is the kind of workflow that saves real time once it is set up.
For Vibe Builders, the appeal is the lowered wall: you can ask for "average order value by region" or spin up an Executive Overview dashboard without learning LookML syntax or hunting through the Looker UI. If your company already has a Looker instance and someone can hand you API credentials, this turns your AI agent into a data analyst you can talk to.
For Developers and data engineers, the Looker Development bundle is the draw -- generating views and measures from prompts and automating semantic-layer maintenance. It fits naturally into a Claude Code, Antigravity, or Codex workflow you already use for code.
Skip it if you do not run Looker, if you need a stable production dependency today (it is beta and can break between releases), or if you want a standalone BI tool rather than an add-on for an existing instance. For those cases, a dedicated analytics product or a generic database connector will serve you better.
Frequently Asked Questions
What is Looker Agent Skills?
Looker Agent Skills is an open-source set of agent skills, published by Google under the Apache-2.0 license, that connects AI agents to a Looker business intelligence instance. It lets you explore data, manage dashboards, and develop LookML using natural language prompts instead of the Looker UI or hand-written code. The project reached version 0.3.5 in June 2026 and works with Antigravity, Claude Code, and Codex.
Is Looker Agent Skills free and open-source?
Yes. Looker Agent Skills is released by Google under the Apache-2.0 license, which makes it free to use and modify. The skills themselves cost nothing, but you do need a paid Looker instance and valid API credentials to connect to your data. As of 2026 the project is in beta (pre-v1.0), so it is free to adopt but not yet positioned as a stable enterprise dependency.
Which AI agents support Looker Agent Skills?
Looker Agent Skills installs into several host agents rather than running on its own. It supports Antigravity CLI and Antigravity 2.0, Claude Code version 2.1.94 or higher through the marketplace add and plugin install flow, and Codex version 0.117.0 or higher. You can also register the skills with the open npx skills command, and each host requires the Looker environment variables to be set first.
What do you need to use Looker Agent Skills?
You need an existing Looker instance and Looker API credentials, specifically a client ID and a client secret, plus the base URL of your instance. These are provided through the LOOKER_BASE_URL, LOOKER_CLIENT_ID, and LOOKER_CLIENT_SECRET environment variables. You also need one of the supported AI agents installed at its minimum version, since the skills run inside the agent rather than as a separate application.
Is Looker Agent Skills production-ready?
Not fully. As of 2026 Looker Agent Skills is in beta (pre-v1.0), and the maintainers explicitly warn that breaking changes may happen until the first stable release. For production use you should pin a specific version, watch the GitHub releases page, and test changes before upgrading. It is a strong choice for experimentation and internal workflows, but treat it as evolving software rather than a locked dependency.
What is looker?
Looker Agent Skills is an open-source plugin by Google that connects AI agents like Claude Code, Antigravity, and Codex to Looker, so you can explore data, manage dashboards, and write LookML using natural language. Think of it as a prompt-driven control layer over your Looker instance.
How do I install looker?
Visit the GitHub repository at https://github.com/gemini-cli-extensions/looker for installation instructions.
What license does looker use?
looker uses the Apache-2.0 license.
What are alternatives to looker?
Explore related tools and alternatives on My AI Guide.
Open source: source code publicly visible
Anyone can inspect exactly what this repo does on GitHub before using it. Apache-2.0 licensed.
Reviewed by My AI Guide for relevance, quality, and active maintenance before listing.
Install in Claude Code:
/install looker