AI IDE

AI-powered code editor built on VS Code with autonomous agents, multi-model support, and team collaboration.

Overview

AI IDE is built on VS Code, the world’s most popular code editor, enhanced with deeply integrated AI capabilities. Beyond simple code completion, it offers autonomous agents, AI councils for collaborative decision-making, and seamless integration with multiple AI providers.

Key Features

AI Chat & Agent Modes

Two powerful modes for different workflows:

Switch between modes instantly based on your task complexity.

Multi-Model Support

Choose from a wide range of AI models:

• Anthropic: Claude Sonnet 4.5, Claude Haiku 4.5, Claude Opus
• Mistral: Codestral, Devstral (optimized for code)
• Google: Gemini models
• OpenAI: GPT-4o, Codex
• Local: Ollama for offline work

Each model shows token limits so you can choose based on context needs.

Context-Aware Assistance

The AI understands your project structure with smart context tags:

• Backend - Server-side code and APIs
• Frontend - UI components and client code
• DevOps - Infrastructure and deployment
• Database - Queries and schema
• Security - Auth and vulnerability concerns
• Testing - Test coverage and quality

Code Actions

Right-click any code for instant AI actions:

Visual Thinking

AI generates diagrams directly in chat:

• Architecture diagrams
• Flowcharts
• Sequence diagrams
• State machines
• ER diagrams

Ask “show me the architecture” and get a visual diagram with explanations.

Task Planning

For complex features, AI creates structured plans:

• Feature priority matrices
• Phase-by-phase breakdowns
• Key todos with checkboxes
• Dependency tracking

Autonomous Agents

The IDE includes a powerful autonomous agent system for hands-off task completion.

One-Shot Tasks

Describe what you want built and let the agent handle it:

1. Open the One-Shot Task panel
2. Describe your goal (e.g., “Build a REST API for user management”)
3. Set completion criteria (tests pass, docs generated, etc.)
4. Choose provider, model, and iterations
5. Click Start Agent

The agent will autonomously:

• Plan the implementation
• Write code across multiple files
• Run tests and fix failures
• Generate documentation
• Stop when criteria are met

Agent Settings

Fine-tune agent behavior:

AI Councils

Councils bring multiple AI perspectives together for better decisions.

What Are Councils?

A council is a group of AI agents that deliberate on a question, each bringing different expertise. They discuss, debate, and reach consensus before providing recommendations.

Preset Councils

Council Configuration

Customize how councils operate:

• Deliberation Mode: Collaborative, Overseer, or Debate
• Max Deliberation Rounds: How many discussion cycles
• Consensus Threshold: Percentage needed for agreement
• Human Approval: Require sign-off before execution

Custom Councils

Create councils tailored to your team:

1. Click + in the Councils panel
2. Name your council and describe its purpose
3. Add members from different sources:
   • Native LLM - Direct model access
   • Agent Engine - Autonomous agents
   • Amplifier Agents - Specialized experts
   • CLI Agent - Claude Code, Codex, Gemini CLI

Amplifier

Amplifier provides 30+ specialized AI agents organized by expertise.

Agent Categories

Analysis & Debugging (4 agents)

• Bug hunters and code analyzers

Architecture & Design (6 agents)

• System architects and design experts

Content & Research (5 agents)

• Documentation and research specialists

Building & Development (7 agents)

• contract-spec-author - API contract design
• integration-specialist - System integration
• modular-builder - Component architecture
• module-intent-architect - Intent-driven design
• performance-optimizer - Speed and efficiency
• test-coverage - Test analysis and generation
• zen-architect - Holistic system design

Knowledge & Synthesis (5 agents)

• Learning and knowledge management

Meta Tools (3 agents)

• amplifier-cli-architect - CLI expertise
• post-task-cleanup - Code cleanup
• subagent-architect - Agent coordination

Knowledge Base

Amplifier maintains a knowledge graph of your codebase:

• Query Knowledge - Ask questions about your code
• Update Knowledge - Refresh the index
• View Graph - Visualize code relationships
• Show Statistics - Coverage and metrics

CLI Integration

Connect to external AI CLIs:

• Claude Code - Anthropic’s coding agent
• Codex - OpenAI’s code model
• Gemini - Google’s multimodal AI

Switch between backends with a single click.

Worktrees

Manage multiple working directories:

• Create Worktree - New isolated workspace
• List All - See all worktrees
• Switch Worktree - Change active workspace

Perfect for working on multiple features or branches simultaneously.

Getting Started

Opening the IDE

1. Go to the Hub
2. Click IDE
3. Your workspace opens with the AI panel ready

Your First AI Task

1. Open any code file
2. Press Cmd/Ctrl + Shift + A to open AI Chat
3. Ask: “Explain this codebase structure”
4. See the AI analyze your project

Running an Autonomous Task

1. Click the Agents icon in the sidebar
2. Select One-Shot tab
3. Describe: “Add authentication to this Express app”
4. Set iterations to 50
5. Click Start Agent
6. Watch as it builds the feature

Keyboard Shortcuts

When to Use IDE

Documentation