OpenBB Widgets JSON MCP provides structured, programmatic access to the OpenBB widgets.json specification for building financial data applications. Rather than manually parsing lengthy documentation, developers can query specific widget types, configuration options, and parameter schemas through dedicated MCP tools. Key capabilities include: - Widget configuration tools for categories, grid sizes, refetch intervals, error handling, and render functions - Widget parameter documentation covering text inputs, date pickers, boolean toggles, dropdowns (including advanced and dependent variants), number inputs, and input forms - Widget type specifications for Plotly charts, JSON data, tables, and other visualization components - JSON schema generation for widgets, applications, and AI agents - Programmatic access to OpenBB developer documentation built with FastMCP

OpenBB Widgets JSON MCP Server

An MCP server that exposes the OpenBB widgets.json specification as structured, callable tools. Instead of parsing long-form documentation, developers (and AI coding assistants like Claude Code) can directly query widget types, inputs, and configuration examples through this server.

What This Does

Each widget type is mapped to its own tool—making it easy to generate, mix, and match valid JSON specs when building new OpenBB apps. The server is built with FastMCP, programmatically generated from the OpenBB developer docs, and hosted on Smithery.ai for discoverability.

The server provides:

Prerequisites

Quick Start

1. Clone the repository:

   git clone https://github.com/OpenBB-finance/openbb-widgets-json-mcp.git
   cd openbb-widgets-json-mcp
   

2. Install dependencies:

   uv sync
   

3. Run the server:

   uv run python main.py
   

4. Test it's working: The server will start on http://localhost:8081

5. Launch Smithery Playground (optional):

   npm install -g @smithery/cli
   smithery playground --port 8081
   

Available MCP Tools

Via iterating through OpenBB docs

#### Widget Configuration Tools

#### Widget Parameter Tools

#### Widget Type Tools

#### JSON Reference Tools

#### Additional Tools

Added ad-hoc

Example: Playground Monte Carlo Widget

The playground/ folder contains a complete Monte Carlo simulation widget built using Claude Code and this MCP server. This demonstrates the full development workflow:

1. Claude Code queried the MCP server for OpenBB widget specifications 2. MCP tools provided boilerplate code, plotly chart documentation, and JSON reference specs 3. Generated widget includes: - 3 parameters: ticker (text), start_date (date), use_volatility_adjustment (boolean) - Plotly visualization with 50 simulation paths + percentile bands - Raw data mode returning pure simulation arrays - Run button for manual execution - Full OpenBB Workspace integration

To run the example:

cd playground
uv sync
uv run python main.py

This serves as a practical example of how AI coding assistants can leverage MCP servers to rapidly