Skip to content

DEMO_SCRIPT.md

Latest commit

 

History

History
116 lines (98 loc) · 3.9 KB

DEMO_SCRIPT.md

File metadata and controls

116 lines (98 loc) · 3.9 KB
Raw

Code Quality Intelligence Agent - Demo Script (5-7 minutes)

Overview

This demo showcases our AI-powered Code Quality Intelligence Agent that analyzes multi-language codebases, identifies quality issues, and provides actionable insights with interactive Q&A.

Demo Flow

1. Introduction (30 seconds)

  • "I built a Code Quality Intelligence Agent for the Atlan AI Engineering internship challenge"
  • "It analyzes Python and JavaScript/TypeScript codebases, finds quality issues, and provides interactive Q&A"
  • "Let me show you the key features: CLI analysis, web UI, and intelligent code search"

2. CLI Analysis (2 minutes)

# Show the CLI in action
python -m cq_agent.cli.main analyze . --md demo-report.md --sarif demo-report.sarif

# Highlight the output:
# - Files: 15 | SLOC: 1255 | Languages: python
# - Issues found: 31 (shows top 10)
# - Generated reports

# Show autofix preview
python -m cq_agent.cli.main analyze . --autofix-dry-run
# - Shows unified diff of safe fixes (unused imports)

Key points to mention:

  • Multi-language support (Python + JS/TS)
  • Prioritized issues by severity and impact
  • SARIF output for IDE/CI integration
  • Safe autofix with dry-run preview

3. Web UI Demo (2.5 minutes)

# Launch the Streamlit UI
streamlit run src/cq_agent/web/app.py

Show in UI:

  1. Overview KPIs: Files, SLOC, Languages, Issues count
  2. Dashboards tab:
    • Severity distribution bar chart
    • Top hotspots bar chart
  3. Issues tab:
    • Filter by severity/category/source
    • Search functionality
    • Color-coded severity badges
  4. File Details tab:
    • Select a file with issues
    • Show code context around issues
    • Expandable issue details
  5. Autofix tab:
    • Preview safe fixes
    • Show confirmation flow
  6. Export tab:
    • Download Markdown/CSV reports

Key points:

  • Interactive filtering and search
  • Visual dashboards with charts
  • Per-file issue details with code context
  • Safe autofix with confirmation
  • Export capabilities

4. Q&A Demo (1.5 minutes)

# Start interactive Q&A
python -m cq_agent.cli.main qa .

# Ask questions like:
# - "How does the ingestion work?"
# - "What are the main analyzers?"
# - "Show me security issues"
# - "How is scoring implemented?"

Key points:

  • Enhanced TF-IDF search with function/filename boosting
  • File:line citations in results
  • Natural language queries over codebase
  • Relevant code snippets with context

5. Architecture & Technical Highlights (1 minute)

  • Modular design: ingestion → analyzers → scoring → reporting
  • Smart prioritization: git churn × complexity × centrality
  • Multi-analyzer: Ruff, Bandit, Radon, ESLint
  • Hotspots detection: dependency graph analysis
  • Safe autofix: deterministic fixes with confirmation
  • CI/CD ready: GitHub Action + SARIF integration

6. Conclusion (30 seconds)

  • "This agent provides comprehensive code quality analysis with practical developer tools"
  • "Key differentiators: smart prioritization, interactive UI, safe autofix, and intelligent Q&A"
  • "Ready for production use with CI/CD integration and extensible architecture"

Demo Tips

  • Keep terminal and browser windows side by side
  • Have sample outputs ready in sample-outputs/
  • Practice the Q&A questions beforehand
  • Show confidence in the technical decisions
  • Emphasize practical developer value

Backup Plan

If something fails:

  • Show the sample outputs in sample-outputs/
  • Explain the architecture from the notebook
  • Highlight the GitHub Action workflow
  • Discuss the modular, extensible design

Key Messages

  1. Practical: Solves real developer problems
  2. Intelligent: Smart prioritization and search
  3. Safe: Confirmation-based autofix
  4. Extensible: Clean modular architecture
  5. Production-ready: CI/CD integration