Skip to main content

Quick Start Guide

UnfoldAI monitors your AI coding agents — Cursor, Claude Code, Copilot, and others — and surfaces errors, drift, and risk signals so you can supervise instead of babysit. Here's how to get started.

Step 1: Open the Dashboard​

Click the UnfoldAI icon in the Activity Bar, or use the keyboard shortcut:

  • Ctrl+Shift+U / Cmd+Shift+U — Open UnfoldAI Terminal
  • Click the UnfoldAI icon in the sidebar to open the panel

The Dashboard is your home base. It shows your workspace health at a glance: agent activity, active issues, and recent events.

Step 2: Run Your AI Agent​

Use your AI coding agent as you normally would — run Cursor, start a Claude Code session, use Copilot, or any other agent. UnfoldAI automatically detects agent activity and begins tracking.

You don't need to configure anything. Agent detection is automatic for all supported agents.

Step 3: Watch the Dashboard Update​

As your agent works, the Dashboard updates in real time:

  • Agent Activity shows which agents are active and what they're doing
  • Activity Feed shows a timeline of commands, file changes, and events
  • Health Status shifts from healthy to warning/error if issues are detected

Step 4: When Something Goes Wrong​

When UnfoldAI detects an error or issue, it creates an Episode — a grouped bundle of related events, errors, and context. Episodes appear as alert cards on the Dashboard and as proactive messages in the chat stream.

From an Episode, you can:

  • Investigate — See the full error context and timeline
  • Chat — Ask UnfoldAI to analyze the issue with pre-loaded context
  • Get a Fix Plan — Request an AI-generated resolution plan
  • Save a Playbook — Store the solution for next time

Step 5: Before You Commit​

Before committing agent-written code, open the Pre-Commit Summary:

  • Command Palette → UnfoldAI: Pre-commit Summary

This shows you everything that changed: files modified, risk signals, dependency changes, and an AI-generated commit message.

What's Next​