Apache Iceberg

Open table format for huge analytic datasets with agent-friendly branching and time travel.

MCPCLI

About Apache Iceberg

Open table format for huge analytic datasets with agent-friendly branching and time travel. Explore how Apache Iceberg integrates with the agentic data stack ecosystem and supports autonomous data operations.

Key Features

  • ACID transactions on cloud object storage ensuring data integrity
  • Schema evolution without table rewrites (add, rename, reorder columns)
  • Hidden partitioning that decouples physical layout from query filters
  • Time travel and version rollback for reproducible queries and recovery
  • Multi-engine compatibility (Spark, Trino, Flink, Presto, Hive, Impala)
  • Partition evolution allowing partition schemes to change without rewriting data
  • Scalable metadata handling for tables with tens of petabytes of data
  • File-level statistics enabling query engines to skip irrelevant data files

Agent Integration

CLI — pyiceberg

$ pip install "pyiceberg[cli]"
CLI Documentation

External Links

Cloudera Iceberg MCP Server

MCP server providing read-only access to Iceberg tables via Impala with LangChain/OpenAI SDK integration

REST Catalog Spec

Official REST API specification for Iceberg catalogs — the standardized API agents use

PyIceberg (Python SDK)

Native Python library for programmatic access to Iceberg table metadata and data, no Spark/JVM required

Awesome Apache Iceberg

Curated list of Apache Iceberg resources, tools, and ecosystem projects

Branching and Tagging Docs

Documentation for git-like branching, tagging, and fast-forward merge on Iceberg tables

← Back to Lake Format