Hudley.ai Agentic Studio Documentation
Stop typing code. Start directing AI agents.
Hudley is an autonomous AI development studio that transforms how you build software. Instead of manually writing every line of code, you describe features on a Kanban board and watch as AI agents powered by Claude Agent SDK automatically implement them.
1 What Makes Hudley Different?
Traditional development tools help you write code. Hudley helps you orchestrate AI agents to build entire features autonomously. Think of it as having a team of AI developers working for you—you define what needs to be built, and Hudley handles the implementation.
The Workflow
Add Features
Describe features you want built (with text, images, or screenshots)
Move to "In Progress"
Hudley automatically assigns an AI agent to implement the feature
Watch It Build
See real-time progress as the agent writes code, runs tests, and makes changes
Review & Verify
Review the changes, run tests, and approve when ready
Ship Faster
Build entire applications in days, not weeks
Powered by Claude Agent SDK
Hudley leverages the Claude Agent SDK to give AI agents full access to your codebase. Agents can read files, write code, execute commands, run tests, and make git commits—all while working in isolated git worktrees to keep your main branch safe.
2 Prerequisites
- Node.js 18+ (tested with Node.js 22)
- npm (comes with Node.js)
-
Authentication (choose one):
- • Claude Code CLI (recommended) - Install and authenticate, credentials used automatically
- • Anthropic API Key - Direct API key for Claude Agent SDK
3 Quick Start
After downloading and installing Hudley, launch the application. On first run, you'll see a setup wizard where you can configure authentication.
Authentication Options:
- • Use Claude Code CLI (recommended) - Hudley will detect your CLI credentials automatically
- • Enter an API key directly in the wizard
4 How to Run
Desktop Application
Simply download the installer for your platform (macOS, Windows, or Linux) and run it. The desktop application provides the best experience with native window management and system integration.
For Developers (Source)
If you're running from source code:
You'll be prompted to choose between Web Application (browser) or Desktop Application (Electron).
5 Authentication Setup
Option 1: Claude Code CLI (Recommended)
Install and authenticate the Claude Code CLI. Once authenticated, Hudley will automatically detect and use your CLI credentials. No additional configuration needed!
Option 2: Direct API Key
If you prefer not to use the CLI, you can provide an Anthropic API key directly using one of these methods:
Environment Variable
.env File
In-App Settings
You can also enter your API key directly in the Settings view within the application.
6 Available Views
Hudley provides several specialized views accessible via the sidebar or keyboard shortcuts:
| View | Shortcut | Description |
|---|---|---|
| Board | K |
Kanban board for managing feature workflow (Backlog → In Progress → Waiting Approval → Verified) |
| Agent | A |
Interactive chat sessions with AI agents for exploratory work and questions |
| Spec | D |
Project specification editor with AI-powered generation and feature suggestions |
| Context | C |
Manage context files (markdown, images) that AI agents automatically reference |
| Profiles | M |
Create and manage AI agent profiles with custom prompts and configurations |
| Settings | S |
Configure themes, shortcuts, defaults, authentication, and more |
| Terminal | T |
Integrated terminal with tabs, splits, and persistent sessions |
| Running Agents | - | View all active agents across projects with status and progress |
7 Keyboard Shortcuts
All shortcuts are customizable in Settings. Default shortcuts:
Navigation
-
Board
K -
Agent
A -
Spec
D -
Context
C -
Settings
S -
Profiles
M -
Terminal
T
Actions
-
Toggle Sidebar
` -
New Item
N -
Start Next Features
G -
Open Project
O -
Project Picker
P -
Previous Project
Q -
Next Project
E
8 Core Workflow Features
Kanban Board
Visual drag-and-drop board to manage features through backlog, in progress, waiting approval, and verified stages
AI Agent Integration
Automatic AI agent assignment to implement features when moved to "In Progress"
Git Worktree Isolation
Each feature executes in isolated git worktrees to protect your main branch
Real-time Streaming
Watch AI agents work in real-time with live tool usage, progress updates, and task completion
Follow-up Instructions
Send additional instructions to running agents without stopping them
9 AI & Planning
Multi-Model Support
Choose from Claude Opus, Sonnet, and Haiku per feature
Extended Thinking
Enable thinking modes (none, medium, deep, ultra) for complex problem-solving
Planning Modes
Four planning levels: skip (direct implementation), lite (quick plan), spec (task breakdown), full (phased execution)
Plan Approval
Review and approve AI-generated plans before implementation begins
Multi-Agent Task Execution
Spec mode spawns dedicated agents per task for focused implementation
10 Project Management
Project Analysis
AI-powered codebase analysis to understand your project structure
Feature Suggestions
AI-generated feature suggestions based on project analysis
Context Management
Add markdown, images, and documentation files that agents automatically reference
Dependency Blocking
Features can depend on other features, enforcing execution order
Graph View
Visualize feature dependencies with interactive graph visualization
11 Developer Tools
Integrated Terminal
Full terminal access with tabs, splits, and persistent sessions
Image Support
Attach screenshots and diagrams to feature descriptions
Concurrent Execution
Configure how many features can run simultaneously (default: 3)
Keyboard Shortcuts
Fully customizable shortcuts for navigation and actions
Theme System
25+ themes including Dark, Light, Dracula, Nord, Catppuccin, and more
Cross-Platform
Desktop app for macOS (x64, arm64), Windows (x64), and Linux (x64)
12 How It Works
Feature Definition
Users create feature cards on the Kanban board with descriptions, images, and configuration
Git Worktree Creation
When a feature starts, a git worktree is created for isolated development
Agent Execution
Claude Agent SDK executes in the worktree with full file system and command access
Real-time Streaming
Agent output streams via WebSocket to the frontend for live monitoring
Plan Approval (optional)
For spec/full planning modes, agents generate plans that require user approval
Verification
Features move to "Waiting Approval" where changes can be reviewed via git diff
Integration
After approval, changes can be committed and PRs created from the worktree
13 Tech Stack
Frontend
- React 19 - UI framework
- Vite 7 - Build tool
- Electron 39 - Desktop framework
- TypeScript 5.9 - Type safety
- TanStack Router - File-based routing
- Zustand 5 - State management
- Tailwind CSS 4 - Styling
- xterm.js - Terminal emulator
Backend
- Node.js - JavaScript runtime
- Express 5 - HTTP server
- TypeScript 5.9 - Type safety
- Claude Agent SDK - AI integration
- WebSocket (ws) - Real-time streaming
- node-pty - Terminal sessions
14 Data Storage
Hudley uses a file-based storage system (no database required):
Per-Project Data
Stored in {projectPath}/.hudley/
.hudley/
├── features/ # Feature JSON files and images
│ └── {featureId}/
│ ├── feature.json # Feature metadata
│ ├── agent-output.md # AI agent output log
│ └── images/ # Attached images
├── context/ # Context files for AI agents
├── settings.json # Project-specific settings
├── spec.md # Project specification
└── analysis.json # Project structure analysis
Global Data
Stored in the application data directory:
data/ ├── settings.json # Global settings, profiles, shortcuts ├── credentials.json # API keys (encrypted) ├── sessions-metadata.json # Chat session metadata └── agent-sessions/ # Conversation histories
15 Security & Isolation
Important Security Note
This software uses AI-powered tooling that has access to your operating system and can read, modify, and delete files. Use at your own risk. We recommend using Hudley within a sandboxed environment or Docker container for maximum security.
Git Worktrees
Each feature executes in an isolated git worktree, protecting your main branch
Path Sandboxing
Optional ALLOWED_ROOT_DIRECTORY restricts file access to specific directories
Docker Isolation
Recommended deployment uses Docker with no host filesystem access
Plan Approval
Optional plan review before implementation prevents unwanted changes