agentic-coding-demo/demo-05/README.md

Demo 05: Codebase Navigation - Understanding Vitabel's Interactive Plotting

Purpose

This demo shows how to use an AI agent to navigate and understand a complex codebase. Instead of reading hundreds of files manually, you'll use the agent to quickly locate relevant code, understand the architecture, and dissect how a specific feature works.

Target Codebase: Vitabel

Vitabel is a Python toolbox for plotting, processing, and interactively labeling vital medical data (e.g., from defibrillators, anesthesia records, or critical care monitors). It's developed by the Uni Graz Math department.

Repository: https://github.com/UniGrazMath/vitabel

Key Stats:

• ~46 MB codebase
• Python-based
• Designed for Jupyter notebooks
• Main feature: Interactive data visualization and annotation

Task: Understand the Interactive Plotting System

Your goal is to understand how the interactive plotting system in Vitabel works. Specifically:

1. Find the entry point: Locate the plot_interactive() method mentioned in the README
2. Trace the flow: Follow the code path from the method call to the actual rendering
3. Identify key components: What libraries are used? What classes are involved?
4. Understand the architecture: How are plots structured? How does interactivity work?
5. Document the findings: Create a brief summary of how the system works

Starting Point

From the Vitabel README, we know the main entry point is:

case.plot_interactive(
    channels=[['cpr_acceleration'], ['capnography'], ['ecg_pads'], []],
    labels = [['ROSC'], ['etco2_from_capnography', 'ROSC'], ['ROSC'], ['ROSC', 'rosc_probability']],
    channel_overviews=[['cpr_acceleration']],
    time_unit='s',
    subplots_kwargs={'figsize': (22, 9)}
)

Your task: Find and understand how this plot_interactive() method is implemented.

Process

Use the agent iteratively to:

1. Explore the codebase structure: What directories exist? What are the main modules?
2. Locate the plotting code: Search for plot_interactive and related plotting functions
3. Understand the dependencies: What libraries (matplotlib, plotly, etc.) are used?
4. Follow the call chain: From the public API to the internal implementation
5. Identify key classes/functions: What are the main building blocks?
6. Ask targeted questions: Use the agent to explain complex parts or unfamiliar patterns

Tips for Agent-Assisted Navigation

• Start broad: Ask the agent to describe the overall structure first
• Be specific: Once you know where to look, ask about specific functions/classes
• Ask for context: Request explanations of unfamiliar patterns or libraries
• Iterate: Use answers to refine your questions and dig deeper
• Focus: Don't try to understand everything - focus on the plotting system

Expected Outcome

By the end of this demo, you should have:

• A mental map of Vitabel's plotting architecture
• Understanding of how plot_interactive() works under the hood
• Experience using an AI agent to navigate and understand a complex codebase

Files to Explore (Suggestions)

Start with these areas:

• src/vitabel/ - Main source code directory
• Look for vitals.py or similar module containing the Vitals class
• Search for plot_interactive implementation
• Check for visualization-related modules or subdirectories

Bonus Challenges

If you finish early, try:

1. Understand how the label system works (how events are annotated on plots)
2. Explore how different data sources are loaded (defibrillator vs VitalDB)
3. Understand the circulation prediction algorithm