Gravitino

High-performance metadata lake that unifies metadata from diverse sources for data and AI.

MCPCLI

About Gravitino

High-performance metadata lake that unifies metadata from diverse sources for data and AI. Explore how Gravitino integrates with the agentic data stack ecosystem and supports autonomous data operations.

Key Features

  • Federated metadata lake acting as a 'Catalog of Catalogs' for unified metadata
  • Multi-catalog support: Hive, Iceberg, MySQL, PostgreSQL, Kafka, Doris, Paimon, Hudi
  • Unified Role-Based Access Control (RBAC) with granular privileges
  • Tag-based data governance for metadata classification and discovery
  • Model Catalog for managing ML models with version tracking and aliases
  • MCP Server integration for connecting AI agents with data context
  • Metadata-driven actions framework for policies and automated job execution
  • Geo-distributed architecture for managing metadata across multiple cloud providers

Agent Integration

CLI — gcli.sh

$ Download binary tarball from GitHub releases and extract
CLI Documentation

External Links

Gravitino MCP Server

Official MCP server with tools for catalogs, schemas, tables, tags, roles, and models

CLI Documentation

CLI reference (gcli.sh) for managing metalakes, catalogs, schemas, tables, tags, and roles

Apache Gravitino GitHub

Main source repo for the federated metadata lake

Building a Universal Data Agent with LlamaIndex + Gravitino

Guide for building AI data agents using Gravitino as the metadata layer

MCP Server Announcement

Blog post explaining MCP server architecture, capabilities, and AI agent use cases

← Back to Catalog Service