Power Tools Guide - Advanced Features

🎯 Overview

Beyond the standard AI chat, Kolega Studio includes powerful specialized agents and advanced tools for complex problem-solving, deep analysis, and sophisticated debugging. These features are accessible across all subscription tiers, with Pro tier users getting additional infrastructure options.

Note: These are advanced features. Most projects won't need them, but when you do, they're incredibly powerful.

🧠 The Multi-Agent System

Kolega Studio uses a sophisticated multi-agent architecture where specialized AI agents work together:

The Agent Team

1. Coding Agent - Your primary conversation partner for building features

2. Investigation Agent - Deep code analysis specialist for understanding existing code

3. Browser Agent - Autonomous testing specialist with screenshot capabilities

4. Thinking Agent - Deep reasoning for complex problems

5. System Agent - Resource coordination across threads

These agents coordinate autonomously, however you can also request specific agents for specialized tasks.

🔍 Investigation Agent

What It Does

The investigation agent provides deep, read-only analysis of your codebase. It systematically explores your project structure, understands relationships between components, and provides architectural insights.

When to Use It

• Before major changes - Understand impact across the codebase

• Debugging complex issues - Trace through code paths

• Learning the codebase - Understand how everything connects

• Gap analysis - Compare implementation plan with existing code

• Security review - Find potential vulnerabilities

• Performance analysis - Identify bottlenecks

How to Activate

In your chat, explicitly request:

"Use the investigation agent to analyze the current codebase structure and explain how the authentication system works" 

or

"Have the investigation agent explore how data flows from the frontend forms to the database" 

🌐 Browser Agent

What It Does

The browser agent autonomously interacts with your application, testing functionality and capturing visual evidence. It's like having a QA tester built into the platform.

When to Use It

• Testing user flows - Verify complete user journeys

• Visual debugging - See exactly what users see with screenshots

• Detecting UI issues - Find layout problems

• Validating interactions - Ensure buttons and forms work

• Screenshot documentation - Capture visual evidence

• Reading error logs - Check console logs for JavaScript errors

How to Activate

Request in chat:

"Use the browser agent to test the complete user registration flow and take screenshots" 

or

"Have the browser agent check if the dashboard displays correctly" 

💭 Think Hard Functionality

What It Does

The "think_hard" capability engages deep analytical reasoning for complex problems. It's like asking Kolega Studio to really consider all options before responding.

When to Use It

• Complex architecture decisions - Choosing between different approaches

• Difficult debugging - When the issue isn't obvious

• Performance optimization - Finding the best solution

• Security analysis - Identifying vulnerabilities

• Algorithm design - Creating efficient solutions

• Mathematical formulas - Complex calculations and implementations

How to Activate

Request explicitly:

"Think hard about the best approach for implementing real-time collaboration in this app" 

or

"Use think hard to figure out why the application slows down when we have more than 100 items" 

🔄 Agent Orchestration

How Agents Work Together

The system intelligently coordinates multiple agents:

1. Primary agent receives your request

2. Dispatches specialized agents as needed

3. Investigation agent analyzes code

4. Browser agent tests functionality

5. Results synthesized into comprehensive response

Automatic Coordination

You'll see this in the Monitor tab:

• dispatch_investigation_agent - Code analysis beginning

• dispatch_browser_agent - Testing starting

• Real-time status updates

• Results aggregation

Manual Agent Requests

You can request multiple agents in one request:

"First, use the investigation agent to analyze the payment system code. Then use the browser agent to test the checkout flow. Finally, think hard about any security vulnerabilities you find." 

🧵 Thread Management

Understanding Threads

Kolega Studio allows multiple threads within a single project - a unique feature that helps with:

• Organization - Keep different features/topics separated

• Token management - Each thread has its own context window

Best Practices for Threads

• New feature = New thread - Start a fresh thread for each major feature

• Context window warnings - When you see warnings about context limits, start a new thread

• Task-focused - Keep threads focused on specific tasks

• Naming threads - Rename threads descriptively for easy navigation

• Testing threads - Use separate threads for testing phases

Thread Context Preservation

When starting a new thread after hitting context limits:

1. Copy the summary from the previous thread

2. Reference the implementation plan document

3. Brief the new thread on what was completed

📊 Monitoring Agent Activity

Watch Agents Work

In the Monitor tab, you'll see:

• "AI Assistant Ready" status

• Real-time agent dispatch notifications

• Tool usage as agents work

• File operations and analysis

• Code generation in progress

• Completion status

Understanding Agent Output

Investigation Agent Output:

Analyzing codebase structure... Found 24 React components Identified 8 API endpoints Detected 3 database models Analysis: The authentication flow uses JWT tokens stored in... 

Browser Agent Output:

Launching browser... Navigating to <http://localhost:9001> Clicking "Login" button Entering credentials Screenshot captured: login_success.png Console log: No errors detected 

🚀 Advanced Workflows

Comprehensive Code Review

"Perform a complete code review: 1. Use the investigation agent to analyze code quality and structure 2. Check for security vulnerabilities 3. Identify performance bottlenecks 4. Use the browser agent to test all user flows 5. Think hard about architectural improvements 6. Provide a detailed report with recommendations" 

Complex Debugging Session

"Debug why users are experiencing slow load times: 1. Use the investigation agent to analyze database queries and API calls 2. Check for N+1 query problems 3. Use the browser agent to measure actual load times 4. Check browser console for errors 5. Think hard about the root cause 6. Propose and implement a fix" 

💡 Pro Tips

Effective Agent Usage

1. Be specific about what you want analyzed

2. Provide context about the problem you're solving

3. Chain agents for comprehensive analysis

4. Watch the Monitor tab to understand what's happening

5. Use new threads when context warnings appear

Common Patterns

Pre-implementation Review:

"Before adding the payment system, use the investigation agent to understand the current order flow" 

Post-implementation Testing:

"After adding the new feature, use the browser agent to test all related user journeys" 

Problem Deep Dive:

"Think hard about why this specific error only occurs after the implementation of our latest feature" 

🎯 Quick Command Reference

🚨 Troubleshooting Agent Issues

Remember: These power tools are for advanced scenarios. Start with standard chat interactions, and bring in specialized agents when you need their unique capabilities. Using threads effectively and calling specific agents directly can significantly improve your development efficiency.