dataio.mcp.server

DataIO MCP Server - Core server implementation.

This server exposes DataIO platform capabilities as MCP tools that can be used by AI assistants like Claude via AWS Bedrock.

Module Contents

Classes

DataIOMCPServer

MCP Server for the DataIO platform.

Functions

get_db_session

Context manager for database sessions.

check_dataset_access

Check if a user has access to a dataset.

Data

logger

API

dataio.mcp.server.logger

‘getLogger(…)’

dataio.mcp.server.get_db_session()[source]

Context manager for database sessions.

dataio.mcp.server.check_dataset_access(session, user_email: str, dataset_id: str) → bool[source]

Check if a user has access to a dataset.

Returns True if:

  • Dataset is public (access_level = DOWNLOAD)

  • User is admin

  • User has direct permission

  • User’s group has permission

class dataio.mcp.server.DataIOMCPServer[source]

MCP Server for the DataIO platform.

Provides tools for:

  • Searching datasets by query, category, tags

  • Getting detailed dataset information

  • Listing categories and data owners

  • Getting download information (for authorized users)

  • Getting dataset schemas/data dictionaries

Initialization

_build_tool_definitions() → dict[str, dataio.mcp.types.BedrockToolSpec][source]

Build the tool definitions for Bedrock.

get_tool_definitions() → list[dict][source]

Get all tool definitions in Bedrock format.

async execute_tool(tool_name: str, tool_input: dict, user_context: dataio.mcp.types.UserContext | None = None) → dataio.mcp.types.ToolResult[source]

Execute a tool with the given input.

Args: tool_name: Name of the tool to execute tool_input: Input parameters for the tool user_context: Optional user context for permission checks

Returns: ToolResult with success status and data or error

async _search_datasets(params: dict, user_context: dataio.mcp.types.UserContext | None) → dataio.mcp.types.ToolResult[source]

Search for datasets.

async _get_dataset_details(params: dict, user_context: dataio.mcp.types.UserContext | None) → dataio.mcp.types.ToolResult[source]

Get detailed dataset information.

async _list_categories() → dataio.mcp.types.ToolResult[source]

List all categories.

async _list_data_owners() → dataio.mcp.types.ToolResult[source]

List all data owners.

async _get_download_info(params: dict, user_context: dataio.mcp.types.UserContext | None) → dataio.mcp.types.ToolResult[source]

Get download information for a dataset.

async _get_dataset_schema(params: dict, user_context: dataio.mcp.types.UserContext | None) → dataio.mcp.types.ToolResult[source]

Get dataset schema/data dictionary.