UI_GUIDE.md

HDF5/NetCDF MCP Server - UI Quick Start Guide

Getting Started

1. Launch the Interface

python app.py

Open your browser to: http://localhost:7860

---

2. Load a File (Required First Step)

The interface has three ways to load a file. All return a file_path that you'll need for every other tool.

Option A: Example Datasets (Easiest)

• Navigate to 📥 File Loading → Example Datasets
• Select a dataset (e.g., small_netcdf)
• Click Submit
• Copy the file_path from the JSON result (e.g., "/tmp/tmpxyz/tiny.nc")

Option B: Download from URL

• Navigate to 📥 File Loading → Download from URL
• Paste a URL to an HDF5 or NetCDF file
• Click Submit
• Copy the file_path from the result

Option C: Upload Local File

• Navigate to 📥 File Loading → Local File
• Select your .nc or .h5 file
• Click Submit
• Copy the file_path from the result

---

3. Use Any Tool

Once you have the file_path, the workflow for all tools is the same:

1. Navigate to the desired tool tab (Data Analysis or Visualizations)
2. Paste the file_path into the "File Path" field
3. Enter the dataset/variable name (e.g., "temperature")
4. Optional: Add a slice string for subsetting (e.g., "0:10, :, :" for first 10 time steps)
5. Click Submit

Example workflow:

1. Load file → Get file_path: "/tmp/tmpxyz/ocean.nc"
2. Structure & Info → Paste file_path → See available variables: ["temperature", "salinity"]
3. Compute Statistics → Paste file_path → Dataset: "temperature" → Slice: "0, :, :"
4. Heatmap → Paste file_path → Dataset: "temperature" → Slice: "0, :, :" → Download HTML

---

Tips

• Keep the file_path handy: Copy it to a text editor - you'll use it repeatedly
• Start with Structure & Info: See what variables are available before analyzing
• Use Preview first: Check a few values before loading full datasets
• Download visualizations: All charts export as interactive HTML files
• Slice large data: Use slice strings (e.g., ":, 0:50, :") to work with subsets