Cryo

Cryo MCP is an API server that exposes Cryo’s blockchain data extraction capabilities via the MCP protocol. You can query blocks, transactions, logs, and more through MCP-compatible clients, then download data and run SQL queries against it. This makes it easy to integrate Cryo-powered blockchain data into your data workflows and applications.

How to use

You interact with Cryo MCP by connecting an MCP client to the Cryo MCP server and issuing high-level data queries. Start by downloading the data you need, then query or filter it with SQL for analysis. Use block ranges, contract filters, and column selection to tailor results to your goals. When you finish, you can retrieve results in JSON, CSV, or Parquet formats and inspect the data schema to understand available columns and data types.

How to install

# Prerequisites
- Python 3.8+
- uv (Ultraviolet) runtime
- A Cryo installation
- An Ethereum RPC endpoint
- DuckDB for SQL query functionality

# Install the Cryo MCP tool via UV
uv tool install cryo-mcp

# Run the MCP server via the interactive Claude workflow
# Step 1: Add an MCP to Claude's prompt
# Step 2: When prompted, choose uvx as the command to run
# Step 3: Provide the runtime command for Cryo MCP
cryo-mcp --rpc-url <ETH_RPC_URL> [--data-dir <DATA_DIR>]

# Optional environment variables
# ETH_RPC_URL for the Ethereum RPC endpoint
# CRYO_DATA_DIR for the local data directory

Configuration and usage notes

Environment variables control where Cryo MCP connects and stores data. The key variables you’ll typically use are ETH_RPC_URL to specify your Ethereum RPC endpoint and CRYO_DATA_DIR to choose the local data directory. You can pass these values directly on the command line or export them in your shell before starting Cryo MCP.

When you start Cryo MCP, you can customize how much data you download and how you access it. Use block range parameters to fetch specific blocks or blocks from the latest, filter by contract addresses, and choose which columns to include. For SQL workflows, always request Parquet format when downloading data for best performance with DuckDB.

Tools and SQL workflow overview

Cryo MCP exposes a set of tools you can call from your MCP client to list datasets, fetch data, inspect schemas, and run SQL queries. The available tooling enables you to explore datasets, download data, and execute SQL against downloaded Parquet files or via on-demand data access.

Security and best practices

Keep your Ethereum RPC URL and any API keys secure. Do not expose the Cryo MCP server to untrusted networks without proper access controls. Use dedicated data directories and restrict network access to the MCP endpoints your clients use.

Troubleshooting tips

If your MCP client cannot connect, verify that Cryo MCP is running with the correct RPC URL and that your environment variables are set as expected. Ensure the Cryo data directory exists and that you have network access to the Ethereum RPC endpoint. If SQL queries return missing columns, consult the dataset schema to confirm available fields for your selected dataset.

Notes and tips

The Cryo MCP server integrates smoothly with MCP-compatible clients. When running SQL queries, aim to download data in Parquet format to leverage DuckDB efficiently. You can query blocks, transactions, logs, and other Cryo datasets, using filters like block ranges and contract addresses to refine results.

Available tools

list_datasets

Returns a list of all available Cryo datasets.

query_dataset

Query a Cryo dataset with filtering options such as blocks, contract, and output format.

lookup_dataset

Get detailed information about a specific dataset, including schema and sample data.

get_latest_ethereum_block

Returns information about the latest Ethereum block.

query_sql

Run a SQL query against downloaded blockchain data, using parquet file paths.

query_blockchain_sql

Query blockchain data with SQL, automatically downloading required data.

list_available_sql_tables

List all SQL-accessible tables that can be queried.

get_sql_table_schema

Retrieve the schema for a specific parquet file.

get_sql_examples

Provide example SQL queries for different datasets.