Documentation
Long Horizon lets your AI coding agent run real browser tests against your application. This documentation covers everything from initial setup to advanced debugging workflows.
Quick start
New to Long Horizon? Follow these steps to get up and running:
- Install the desktop app on your machine
- Connect your first project and configure your app URL
- Enable MCP in your coding agent (Cursor, Claude Code, or Codex)
- Ask your agent to run a test and watch Long Horizon handle the browser automation
Core concepts
Workspaces
Team boundaries for organizing projects, reports, and access control. Learn more →
Projects
Local repositories linked to Long Horizon with test configuration. Learn more →
MCP Server
The bridge between your AI agent and Long Horizon's browser automation. Learn more →
Reports
Shareable test results with full execution history and debugging tools. Learn more →
All topics
- Installation
Get Long Horizon running on macOS or Windows in under five minutes.
- Getting Started
Connect your first project and run your first AI-driven browser test.
- Workspaces
Organize teams, control access, and manage shared test reports.
- Projects
Link local repositories to Long Horizon and configure your test environment.
- MCP Server
Enable your AI coding agent to run browser tests through Model Context Protocol.
- Skills & Subagents
Extend your agent with testing-specific prompts, rules, and automation bundles.
- Reports
Share test results with your team or publish them publicly.
- Local Mode
Store test sessions on your device and understand local-mode feature tradeoffs.
- UI Feedback
Annotate your UI with comments and let AI agents implement the changes.
- Debugging Test Runs
Step through tests, slow down execution, and jump to specific failure points.
- Best Practices
Production-ready patterns for reliable, secure, and maintainable test automation.