Frequently Asked Questions (FAQ) about UBOS MCP Server

Q: What is an MCP Server? A: MCP (Model Context Protocol) Server acts as a bridge, allowing AI models to access and interact with external data sources and tools.

Q: What types of data can the UBOS MCP Server process? A: The UBOS MCP Server is specifically designed to preprocess and analyze CSV files, making it ideal for handling structured data commonly found in spreadsheets and databases.

Q: What are the key features of the UBOS MCP Server? A: The UBOS MCP Server offers a range of features, including data loading and management, data preprocessing (handling null values, mixed data types, etc.), data analysis (statistical summaries, correlation matrices), and data visualization (various plot types). It also supports custom code execution for advanced tasks.

Q: How do I install the UBOS MCP Server? A: You can install the UBOS MCP Server using either uv (recommended) or pip. The installation steps involve adding the mcp[cli] package along with necessary dependencies like pandas, matplotlib, seaborn, and numpy, followed by installing the server script using mcp install server.py.

Q: How do I start the UBOS MCP Server? A: After installation, you can start the MCP Server by running uv run mcp in your terminal.

Q: How do I test the UBOS MCP Server? A: You can test the server using MCP Inspector by running mcp dev server.py. Alternatively, you can install the server in Claude Desktop and interact with it directly.

Q: What is a working directory in the context of the UBOS MCP Server? A: The working directory is the location where the MCP Server reads CSV files from and saves processed data and visualizations to. You can set a new working directory using the set_work_dir(new_work_dir) command.

Q: What kind of data preprocessing can I do with the UBOS MCP Server? A: The server allows you to handle mixed data types in columns, manage null values using various strategies (remove rows, fill with mean/median/mode, forward/backward fill, constant values), drop and rename columns, and run custom dataframe editing code.

Q: What types of data analysis can I perform with the UBOS MCP Server? A: The server provides tools for generating comprehensive data descriptions (statistical summaries), creating correlation matrices with visualizations, and running custom analysis code.

Q: What types of visualizations does the UBOS MCP Server support? A: The server supports various plot types, including line plots, bar charts, scatter plots, histograms with KDE, box plots, violin plots, pie charts, count plots, and kernel density estimation plots. You can also generate custom graphs through code.

Q: Can I run custom code with the UBOS MCP Server? A: Yes, the server allows you to execute custom dataframe manipulation code for preprocessing and custom visualization code for generating graphs.

Q: What is the purpose of the WORK_DIR environment variable? A: The WORK_DIR environment variable specifies the working directory where the UBOS MCP Server reads files from and saves files to.

Q: What error handling does the UBOS MCP Server provide? A: The server includes comprehensive error handling for missing working directories, file not found errors, data loading and processing errors, invalid operations on empty dataframes, mixed data type handling, custom code execution errors, invalid column names, invalid graph types, and null value handling errors.

Q: How does the UBOS MCP Server integrate with the UBOS platform? A: The MCP Server seamlessly integrates with the UBOS platform, allowing you to build AI Agents that can automatically process and analyze data, automate data analysis workflows, and connect to your enterprise data sources.