AI Summit 2025 CLI Demo

A meta-demonstration project for AI Summit 2025: an LLM-powered CLI tool that helps present a talk about building LLM CLI applications.

🎯 The Meta Concept

This project demonstrates itself being built. During the presentation:

• The CLI tool explores its own codebase
• It performs git operations to show incremental development
• It explains its own architecture and implementation
• Each git branch represents a step in building the tool

Think of it as: "Using an LLM CLI to demonstrate how to build an LLM CLI" 🤯

🛠️ Tech Stack

• Python 3.13 - Latest Python with modern features
• Poetry - Dependency management and packaging
• Pydantic AI - LLM interaction and agent framework with tools
• Textual + Rich - Terminal UI and rich text formatting
• Docker - Sandboxed execution environment

🚀 Quick Start

Prerequisites

• Python 3.13+
• Poetry 1.8+
• Docker (optional, for sandboxed execution)

Installation

# Install dependencies
poetry install

# Run the CLI
poetry run aisummit

# Or use the shorter command
poetry run python -m aisummit_cli

Docker Usage

# Build the image
docker compose build

# Run interactive demo shell (recommended)
docker compose run demo
# Inside: aisummit, git, java, mvn available system-wide

# Run single query
docker compose run cli "your query here"

# Development mode with live code mounting
docker compose run cli-dev

Note: Create a .env file with your ANTHROPIC_API_KEY before running.

📚 Demo Branch Structure

The repository is organized into incremental branches, each adding new capabilities:

Branch	Description	Features Added
step/step-0	Initial setup	Poetry config, project structure, Docker
step/step-1	Agent foundation	Pydantic AI agent, streaming responses, conversation history
step/step-2	Rich TUI	Three-section Textual interface, real-time streaming, markdown rendering
step/step-3	Git tools	git_status, git_branch_list, git_log, git_diff, git_checkout
step/step-4	File tools	glob_files, grep_content, read_file
step/step-5	Edit tools	replace_text for file editing

Checking Out Demo Steps

# See current step
git branch

# Jump to a specific step
git checkout step/step-2

# Run the tool at that step
poetry run aisummit

🧰 Tools Overview

Git Tools (step/step-3)

• git_status() - Check working tree status
• git_branch_list() - List all branches
• git_log(count, oneline) - View commit history
• git_diff(file, compare_to) - Show code changes
• git_checkout(branch_or_tag) - Switch between branches

File Tools (step/step-4)

• glob_files(pattern) - Find files matching patterns
• grep_content(pattern, path) - Search file contents with regex
• read_file(path, start_line, end_line) - Display file contents

Edit Tools (step/step-5)

• replace_text(file_path, old_string, new_string) - Edit files with string replacement

🎓 Conference Demo Flow

1. Introduction - Show step/step-0, explain the meta concept
2. Agent Setup - Checkout step/step-1, demonstrate LLM integration with streaming
3. UI Enhancement - Checkout step/step-2, show rich TUI with real-time streaming
4. Git Tools - Checkout step/step-3, use the tool to explore its own git history
5. File Tools - Checkout step/step-4, search and read the tool's own code
6. Edit Tools - Checkout step/step-5, make changes with replace_text tool

🔒 Safety & Sandboxing

• All operations run in Docker with limited permissions
• Git operations are read-only (no commits/pushes during demo)
• File operations restricted to repository directory
• No system shell access exposed to the LLM

📖 Documentation

See CLAUDE.md for detailed architecture and development guidance.

🤝 Contributing

This is a conference demo project. Feel free to:

• Fork and adapt for your own presentations
• Suggest improvements via issues
• Share your own LLM CLI implementations

📝 License

MIT License - See LICENSE file for details

---

Built for AI Summit 2025 | Where the demo demonstrates itself ✨