Pi AI Agent Toolkit Overview | Generated by AI

Home 2026.02

I’ll fetch and analyze that GitHub repository for you.Let me fetch the detailed README and AGENTS documentation to give you a more comprehensive overview.# Pi Monorepo - Comprehensive Introduction

Pi-mono is an AI agent toolkit created by Mario Zechner (badlogic) that provides a complete ecosystem for building and deploying AI coding agents and managing LLM interactions. It’s structured as a monorepo containing multiple interconnected packages.

Core Philosophy

Pi takes a minimalist, extensible approach. Rather than shipping with every possible feature, it provides powerful defaults and lets you build what you need through skills, extensions, prompt templates, or third-party packages. The focus is on flexibility and real-world usability.

Main Packages

1. @mariozechner/pi-coding-agent (The Flagship)

An interactive coding agent CLI that runs in your terminal. This is the main product most users interact with.

Key Features:

Basic Usage:

# Install
npm install -g @mariozechner/pi-coding-agent

# Interactive mode
pi "List all .ts files in src/"

# Non-interactive
pi -p "Summarize this codebase"

# Specify model
pi --provider openai --model gpt-4o "Help me refactor"

# Read-only mode
pi --tools read,grep,find,ls -p "Review the code"

2. @mariozechner/pi-ai

The foundational LLM toolkit providing a unified API across multiple providers.

Supported Providers:

Key Features:

3. @mariozechner/pi-agent

Core agent runtime providing state management and tool execution orchestration.

Capabilities:

4. @mariozechner/pi-tui

A terminal UI library with differential rendering for building responsive CLI interfaces.

Features:

5. @mariozechner/pi-web-ui

Web components for building AI chat interfaces in browsers.

6. @mariozechner/pi-mom

A Slack bot that delegates messages to the pi coding agent, enabling team collaboration through Slack.

7. @mariozechner/pi-proxy

CORS proxy for making browser-based LLM API calls without exposing keys (though production apps should use proper backends).

8. @mariozechner/pi (pi-pods)

CLI tool for managing vLLM deployments on GPU pods, useful for self-hosting models.

Extensibility System

Skills

Prompt templates or instructions that enhance the agent’s capabilities. They’re loaded and injected into the system prompt.

Extensions

TypeScript/JavaScript code that hooks into the agent’s lifecycle through events:

Extensions can:

Pi Packages

Shareable bundles distributed via npm or git containing extensions, skills, prompts, and themes.

# Install from npm
pi install npm:@foo/pi-tools

# Install from git
pi install git:github.com/user/repo

# Local development
pi install /path/to/local/package

Security Note: Extensions run with full system permissions and can execute arbitrary code. Only install from trusted sources.

Configuration & Customization

Model Configuration

Add custom models via ~/.pi/agent/models.json:

{
  "providers": {
    "ollama": {
      "baseUrl": "http://localhost:11434/v1",
      "api": "openai-completions",
      "apiKey": "ollama",
      "models": [
        { "id": "llama3.1:8b" },
        { "id": "qwen2.5-coder:7b" }
      ]
    }
  }
}

Authentication

Models require authentication via:

Settings

Configure behavior through commands:

Environment Variables

Development Workflow

The monorepo uses a lockstep versioning system where all packages maintain the same version number.

# Setup
npm install
npm run build

# Development
npm run dev              # Watch mode for all packages
./pi-test.sh            # Run from source

# Testing
./test.sh               # Non-LLM tests (no API keys needed)
npm test                # All tests

# Version management
npm run version:patch   # 0.7.5 → 0.7.6
npm run version:minor   # 0.7.5 → 0.8.0
npm run version:major   # 0.7.5 → 1.0.0

# Publishing
npm run publish         # Publish all packages to npm

SDK Usage

Embed pi in your own applications:

import { createPiSession } from '@mariozechner/pi-coding-agent';

const session = await createPiSession({
  cwd: process.cwd(),
  model: { provider: 'anthropic', id: 'claude-opus-4-5' }
});

// Send message and handle streaming
await session.sendMessage('Help me refactor this function', {
  onText: (text) => console.log(text),
  onToolCall: (tool) => console.log(`Using tool: ${tool.name}`)
});

// Interrupt or follow up
await session.steer('New instruction');
await session.followUp('After you\'re done, also do this');

Platform Support

Real-World Usage

Pi is actively used in production environments. See openclaw/openclaw for a real-world SDK integration example.

Community & Resources

Why Pi?

Pi stands out for its:

Whether you need a powerful coding assistant, want to build custom AI tools, or manage LLM deployments, pi-mono provides the building blocks.

Back Donate