gemini-cli-extensions/knowledge-catalog
claude-codeSkills for Knowledge Catalog
Knowledge Catalog is an open-source agent extension by Google that connects AI coding agents to Google Cloud Dataplex (formerly Dataplex) for data discovery and governance. Ask in plain English to search entries, look up metadata, and run ad-hoc analytics.
Our Review
Released by Google under the Apache-2.0 license, these agent skills wire Antigravity, Claude Code, and Codex directly into Google Cloud Dataplex (renamed Knowledge Catalog). The single most useful idea here: you describe what you want in plain English and the agent runs the catalog lookups for you, instead of memorizing Dataplex commands.
What Knowledge Catalog adds to your AI agent:
- •Natural-language catalog discovery search entries, look up metadata, and explore aspect types by typing requests like "find all catalog entries related to customer orders".
- •Ad-hoc analytics over governed data ask the agent to calculate metrics directly against the assets your catalog tracks, such as "calculate total customer orders this month".
- •Cross-dataset comparison surface columns that look similar across separate datasets to spot duplication or governance gaps.
- •The Knowledge Catalog Discovery skill the shipped skill that handles entry search, metadata lookup, and aspect-type exploration as one packaged capability.
- •MCP Toolbox under the hood the extension downloads and runs the MCP Toolbox binary so the agent talks to Dataplex over the Model Context Protocol.
Compatibility:
Works with Antigravity (CLI v1.6.0+ or Antigravity 2.0 v2.0.0+), Claude Code v2.1.94+, and Codex v0.117.0+. Install through the Claude Code plugin marketplace, by cloning the repo, or with npx skills add. It needs a Google Cloud project with the Dataplex API enabled, Application Default Credentials in your environment, and the Dataplex Data Reader plus Service Usage Consumer IAM roles.
Limitations:
This project is beta (pre-v1.0, currently v0.5.1) and the maintainers warn of possible breaking changes before the first stable release, so pin a version tag rather than tracking main. It only governs Google Cloud Dataplex data, not other clouds or warehouses, and assumes a working GCP project with the right IAM roles already in place. Adoption is early (13 GitHub stars), so expect thinner community support than mature data tools.
Our Verdict
The knowledge-catalog extension is a focused, official way to reach Google Cloud Dataplex (now branded Knowledge Catalog) governance from inside an AI agent, and it earns a 7.5/10 for doing one job cleanly while staying honest about its beta status.
For Vibe Builders already on Google Cloud, the natural-language layer is the draw: you can search a catalog, check who owns a sensitive table, or pull a quick metric without learning Dataplex syntax. The catch is the setup -- you still need a GCP project, credentials, and IAM roles before the plain-English part pays off.
For Developers and data engineers, the value is keeping catalog work in the same terminal as your code. The MCP Toolbox backing means the agent talks to Dataplex over a standard protocol, and the Apache-2.0 license lets you read and extend the skills. Pin to a release tag like 0.5.1 because main can change under you during beta.
Skip it if your data lives outside Google Cloud, if you need a stable production-grade catalog client today, or if you want a GUI rather than an agent-driven workflow. In those cases a dedicated Dataplex console session or a multi-cloud governance tool fits better.
Frequently Asked Questions
What is the Knowledge Catalog extension?
Knowledge Catalog is an open-source set of agent skills from Google that lets AI coding agents interact with Google Cloud Dataplex using natural language. You can search catalog entries, look up metadata, explore aspect types, and run ad-hoc analytics by typing plain-English requests instead of Dataplex commands. It works with Antigravity, Claude Code, and Codex.
Is Knowledge Catalog free and open source?
Yes. The Knowledge Catalog skills are released by Google under the permissive Apache-2.0 license, so you can use, read, and modify the code freely. The skills themselves are free, but they call Google Cloud Dataplex, so standard Google Cloud usage and API charges still apply to your own project. As of 2026 the project sits at version 0.5.1.
Is Knowledge Catalog the same as Dataplex?
Knowledge Catalog is the new name for the product previously called Dataplex on Google Cloud. The repository name and skills reflect that 2026 rebrand, while the underlying API is still the Dataplex API you enable in your GCP project. So you are governing the same Dataplex data, just under the Knowledge Catalog name.
Who should use Knowledge Catalog and what do I need first?
It suits data engineers, platform engineers, and AI builders who already run data on Google Cloud and want catalog access from their agent. Before installing you need a GCP project with the Dataplex API enabled, Application Default Credentials configured, and two IAM roles: Dataplex Data Reader and Service Usage Consumer. Without that GCP setup the skills cannot reach your catalog.
How do I install Knowledge Catalog in Claude Code?
Set the DATAPLEX_PROJECT environment variable to your GCP project ID, start Claude Code, then add the marketplace with the plugin command pointing at the repo at tag 0.5.1. After that run the plugin install command for knowledge-catalog and verify it with the plugin list command. You can also install via git clone or the npx skills tool for Antigravity and Codex.
What is knowledge-catalog?
Knowledge Catalog is an open-source agent extension by Google that connects AI coding agents to Google Cloud Dataplex (formerly Dataplex) for data discovery and governance. Ask in plain English to search entries, look up metadata, and run ad-hoc analytics.
What license does knowledge-catalog use?
knowledge-catalog uses the Apache-2.0 license.
What are alternatives to knowledge-catalog?
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 knowledge-catalog