KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #codex #typescript #developer-tools #ai-agents #deepseek-harness #claude-code #ai-agent #cli #react #mcp

copilot-studio-agent-optimizer

★ 44 Open GitHub ↗

AI-powered optimization tool for Microsoft Copilot Studio agents. Automatically improves agent configurations with intelligent suggestions and code generation using GitHub Copilot.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Optimize Copilot Studio Agents

![VS Code Extensions Setup](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/banner.png)

**Curated Best Practices Instructions for GitHub Copilot Chat**

### How It Works
1. **Add a prompt file** (best practices instructions) to GitHub Copilot Chat context
2. **Add your YAML file** to the same context  
3. **Ask for optimization** - Copilot now knows exactly what to look for and how to improve it
4. **Get Microsoft-aligned suggestions** based on official documentation

**Transform trigger phrases from 2-3 basic phrases to 5-10 natural variations, enhance conversation quality, and ensure production-ready agent configurations - all through guided AI optimization.**

## 🎯 What is This?

This repository contains **specialized prompt files** that teach GitHub Copilot Chat how to optimize your Microsoft Copilot Studio `.mcs.yml` files according to official Microsoft best practices.

**Simply put:** These are instruction sets that make GitHub Copilot an expert at optimizing your Copilot Studio YAML configurations.

### How It Works
1. **Add a prompt file** (best practices instructions) to GitHub Copilot Chat context
2. **Add your YAML file** to the same context  
3. **Ask for optimization** - Copilot now knows exactly what to look for and how to improve it
4. **Get Microsoft-aligned suggestions** based on official documentation

**Transform trigger phrases from 2-3 basic phrases to 5-10 natural variations, enhance conversation quality, and ensure production-ready agent configurations - all through guided AI optimization.**

### Key Benefits
- **Instant expertise** - Turn GitHub Copilot into a Copilot Studio expert
- **Microsoft-aligned** - All instructions based on official best practices documentation
- **VS Code integrated** - Works seamlessly with GitHub Copilot Chat in your existing workflow
- **Component-specific** - Tailored instruction sets for each Copilot Studio component type
- **No new tools** - Uses GitHub Copilot you already have

## Quick Start Guide

**Follow the steps mentioned below**

### 0️⃣ **Prerequisites**
Before you begin, ensure you have:

1. **Visual Studio Code** installed
2. **Microsoft Copilot Studio Extension** for VS Code
   - Install from the [VS Code Marketplace](https://marketplace.visualstudio.com/items?itemName=ms-copilotstudio.vscode-copilotstudio)
   - Or search "Copilot Studio" in VS Code Extensions (`Ctrl+Shift+X`)
   - **⚠️ Note:** The Copilot Studio Extension is currently in pre-release. Use with caution and test thoroughly in non-production environments.
3. **GitHub Copilot** (free tier available, works with GPT-4o, Claude Sonnet, and other AI models)

![VS Code Extensions Setup](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/vscode-extensions-setup.png)

### 1️⃣ **Get This Repository**
First, you need to download this repository to access the prompt files:

#### **Option A: Clone with Git (Recommended)**
```bash
git clone https://github.com/ramakrishnan24689/copilot-studio-agent-optimizer.git
cd copilot-studio-agent-optimizer
```

#### **Option B: Download ZIP**
1. **Visit the GitHub repository page**
2. **Click "Code" → "Download ZIP"**
3. **Extract to your desired location**

#### **Option C: Open in VS Code directly**
```bash
# Clone and open in VS Code in one command
git clone https://github.com/ramakrishnan24689/copilot-studio-agent-optimizer.git && code copilot-studio-agent-optimizer
```

### 2️⃣ **Choose Your Component**
Pick the prompt file that matches your YAML component:
- `agent.mcs.yml` → Use [AgentConfiguration_BestPractices.md](./prompts/AgentConfiguration_BestPractices.md)
- `CustomTopic.mcs.yml` → Use [CustomTopics_BestPractices.md](./prompts/CustomTopics_BestPractices.md)
- `Fallback.mcs.yml` → Use [SystemTopics_BestPractices.md](./prompts/SystemTopics_BestPractices.md)
- `Search.mcs.yml` → Use [Knowledge_BestPractices.md](./prompts/Knowledge_BestPractices.md)
- And so on... ([see full reference table](#component-quick-reference))

### 3️⃣ **Connect Your Copilot Studio Agent**

To use the AI-powered optimization, connect to your Copilot Studio agent directly from VS Code:

1. **Open VS Code** with this repository 
2. **Open Copilot Studio extension** (from the sidebar or Ctrl+Shift+P → "Copilot Studio")
3. **Select your environment** from the dropdown
4. **Choose your agent** from the available agents in that environment
5. **Clone your agent** to a local subdirectory(/cloned-agents) using the extension's clone option

![Clone Agent](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/copilot-studio-extension-clone-agent.png)

6. **Agent's YAML files will be accessible** in the VS Code workspace for optimization
![MCS files](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/clone-agent-sub-directory.png)

This approach keeps everything in one workspace, making it easy for GitHub Copilot to access both the optimization prompts and your agent's YAML files.

### 4️⃣ **Optimize Your YAML Files**

1. **Ensure your agent is connected** in the Copilot Studio extension
2. **Open Copilot Chat** (Ctrl+Shift+I or click the chat icon)  
3. **Click "Add context"** to select files:
   - Add the relevant prompt file from `prompts/` folder
   - Add your agent's YAML file from the connected workspace
4. **Ask for optimization** using both contexts

**Example Command:**
```
Using the Custom Topics Best Practices as context, evaluate the given [Your custom topic name here] YAML and modify it as needed.
```

>![Simplified workflow](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/copilot-studio-prompt.gif)

### 5️⃣ **See the Transformation**

#### Key Improvements Demonstrated:
- **Trigger Phrases:** Expanded from 2-3 basic phrases to 5-10 natural variations
- **Conversation Design:** Enhanced from simple Q&A to engaging dialogue
- **Entity Integration:** Added proper slot filling and entity validation
- **Error Handling:** Improved fallback responses and user guidance
- **Response Quality:** More conversational and contextually appropriate responses

#### End User Conversation Quality

**Before/After Optimization - Interaction:**
![User Experience Before](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/before-after-conversation.png)

Changed from basic to more natural conversation

#### End User Conversation Quality

Agent instructions modified based on the description and title.

**Agent Overview - Before**
![Agent Instruction Before](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/agent-settings-before.png)

**Agent Overview - After**
![Agent Instruction After](https://raw.githubusercontent.com/Ramakrishnan24689/copilot-studio-agent-optimizer/main/images/agent-settings-after.png)

#### Specific Improvement Example (RegularTopic):

**Topic Recognition:**
- ❌ **Before:** Limited trigger phrases may miss user intents: "I want to check my order status", "Check order status"
- ✅ **After:** Comprehensive trigger coverage with 5-10 varied phrases: "Where's my order?", "Can you help me track my package?", "My delivery status please", "What's the status of my shipment?", "Track my order"

**Response Quality:**
- ❌ **Before:** "Your order status is: [order_status]"
- ✅ **After:** "I found your order! Here's the current status: [order_status]. Would you like me to explain what this means or help you with anything else regarding your order?"

#### Developer Benefits:
- **5-10 minutes** to optimize each topic (vs hours of manual review)
- **Better user satisfaction** through improved conversation quality
- **Higher intent recognition** rates with optimized trigger phrases

---

## 📋 Available Prompts

### 1. [System Topics](./prompts/SystemTopics_BestPractices.md)
**Use for:** Fallback, Greeting, Goodbye, ConversationStart, EndofConversation, etc.
- System trigger configuration
- Conversation flow logic
- User experience design
- Error handling patterns

### 2. [Regular Topics](./prompts/CustomTopics_BestPractices.md)
**Use for:** Custom business logic topics with trigger phrase optimization
- Topic structure and organization
- Trigger phrase optimization (5-10 varied phrases)
- Conversational design
- Entity usage and slot filling
- Natural language understanding
- Integration with other components

### 3. [Knowledge & Search](./prompts/Knowledge_BestPractices.md)
**Use for:** Custom business logic topics
- Topic structure and organization
- Conversational design
- Entity usage and slot filling
- Integration with other components

### 4. [Actions & Power Automate](./prompts/Actions_BestPractices.md)
**Use for:** Search.mcs.yml and knowledge integration
- SearchAndSummarizeContent configuration
- Knowledge source integration
- Response quality optimization
- Performance considerations

### 5. [Bot Settings & Configuration](./prompts/BotSettings_BestPractices.md)
**Use for:** External service integrations and automations
- Connection configuration
- Input/output mapping
- Error handling and security
- User experience during execution

### 6. [Conversation Management](./prompts/ConversationManagement_BestPractices.md)
**Use for:** settings.mcs.yml and bot-level configuration
- Authentication and security
- AI settings optimization
- Feature enablement
- Performance configuration

### 7. [Agent Configuration](./prompts/AgentConfiguration_BestPractices.md)
**Use for:** StartOver, Reset, state management topics
- User confirmation flows
- State management
- Variable cleanup
- Dialog control

### 8. [Entities & Data Types](./prompts/Entities_BestPractices.md)
**Use for:** agent.mcs.yml and GPT agent behavior configuration
- Agent instructions and behavior
- Conversation starters optimization
- GPT capabilities configuration
- Agent personality and tone

- Entity type selection and configuration
- Value definitions and synonyms
- Smart matching optimization
- Slot filling integration

## How to Use These Prompts

### The Core Concept: Teaching GitHub Copilot About Copilot Studio

These prompt files are **instruction sets** that teach GitHub Copilot Chat how to properly optimize Microsoft Copilot Studio YAML files. Think of them as:

- **Expert knowledge transfer** - Each prompt contains Microsoft's official best practices
- **Context for AI optimization** - Tells Copilot exactly what to look for in your YAML
- **Quality standards** - Defines what "good" looks like for each component type
- **Guided improvements** - Provides specific criteria for optimization suggestions

### Intended Workflow with GitHub Copilot Chat

1. **Open GitHub Copilot Chat** in VS Code (Ctrl+Shift+I)
2. **Add context files**:
   - **Instruction file**: One of the prompt files from `prompts/` folder (teaches Copilot the best practices)
   - **YAML file**: Your actual Copilot Studio configuration file (what needs to be optimized)
3. **Ask for optimization**: "Using the best practices as context, evaluate my YAML and suggest improvements"
4. **Get expert-level feedback** based on Microsoft's official guidelines

**Example:**
```
Context Files Added:
✅ prompts/CustomTopics_BestPractices.md (instruction set)
✅ MyOrderTopic.mcs.yml (file to optimize)

Prompt: "Using the Custom Topics Best Practices as context, evaluate my topic YAML and optimize it."

Result: GitHub Copilot optimizes your topic against Microsoft's best practices and suggests specific improvements.
```

#### Recommended Setup:
1. **Visual Studio Code** with Copilot Studio Extensions installed
2. **GitHub Copilot** with your preferred AI model (GPT-4o, Claude Sonnet, etc.)
3. **Connect to your Copilot Studio environment** using the VS Code extension

#### Simple Workflow:
1. **Clone this repository** and open in VS Code
2. **Connect to your Copilot Studio agent**:
   - Open Copilot Studio extension in VS Code
   - Select your environment from the dropdown
   - Choose your agent from the available list
   - Clone the agent to a local subdirectory using the extension
3. **Open Copilot Chat** (Ctrl+Shift+I)
4. **Add context sources** using the "Add context" button:
   - Select the appropriate prompt file from `prompts/` folder
   - Add your agent's YAML files from the cloned subdirectory
5. **Ask for optimization** with both contexts available

#### Example Workflow:
```
1. Clone this repository
2. code . (opens VS Code)
3. Open Copilot Studio extension → Select environment → Choose agent → Clone agent
4. Open Copilot Chat (Ctrl+Shift+I)
5. Add context: prompts/CustomTopics_BestPractices.md
6. Add context: YourTopicName.mcs.yml (from cloned agent subdirectory)
7. Ask: "Using the best practices prompt, optimize my topic and suggest improvements"
```

### Why This Approach Works

- **No new tools needed** - Uses GitHub Copilot you already have
- **Expert-level optimization** - AI gets the same knowledge a Microsoft consultant would use
- **Consistent quality** - Same standards applied across all team members
- **Real-time feedback** - Get suggestions while you're actively editing

## Why Use These Prompts with GitHub Copilot?

### **Turn GitHub Copilot into a Copilot Studio Expert**
- **Knowledge injection** - Each prompt file contains official Microsoft best practices
- **Context-aware optimization** - AI understands your specific YAML structure and provides targeted suggestions
- **Consistent standards** - Ensures all team members get the same expert-level guidance

### **Seamless Integration with Your Workflow**
- **No context switching** - Everything happens in VS Code with tools you already use
- **Real-time assistance** - Get feedback while editing, not after deployment
- **Project-aware** - AI considers your entire bot configuration, not just individual files

### **Quality at Scale**
- **Microsoft-aligned** - Based on official documentation, not opinions
- **Repeatable process** - Same high-quality optimization every time
- **Team consistency** - Everyone gets expert-level guidance regardless of experience

**The key insight: Instead of learning all the best practices yourself, teach them to GitHub Copilot once and get expert optimization on demand.**

## Based on Official Microsoft Guidelines

All prompts are created based on official Microsoft Learn documentation:
- [Topic Authoring Best Practices](https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/topic-authoring-best-practices)
- [Trigger Phrases Best Practices](https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/trigger-phrases-best-practices)
- [Fallback Topic Guidelines](https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/fallback-topic)
- [System Topics](https://learn.microsoft.com/en-us/microsoft-copilot-studio/authoring-system-topics)
- [Configuration Fundamentals](https://learn.microsoft.com/en-us/microsoft-copilot-studio/configuration-fundamentals)

## Important Notes

- **Integrate with VS Code workflow** - Use these prompts within your existing Copilot Studio development environment
- **Leverage AI assistance** - GitHub Copilot's AI models provide context-aware suggestions based on your project structure
- **Preserve YAML structure** - Prompts focus on improvements within existing framework
- **Maintain IDs and references** - Keep topic and action connections intact
- **Follow security guidelines** - Ensure proper authentication and access control
- **Test changes thoroughly** - Validate improvements in test environment first
- **Use VS Code extensions** - Take advantage of YAML validation and Copilot Studio tooling

## Component Quick Reference

| Component Type | File Pattern | Prompt File |
|---|---|---|
| System Topics | `Fallback.mcs.yml`, `Greeting.mcs.yml` | [SystemTopics_BestPractices.md](./prompts/SystemTopics_BestPractices.md) |
| Regular Topics | `EnquireAbout*.mcs.yml`, custom topics | [CustomTopics_BestPractices.md](./prompts/CustomTopics_BestPractices.md) |
| Knowledge | `Search.mcs.yml` | [Knowledge_BestPractices.md](./prompts/Knowledge_BestPractices.md) |
| Actions | `actions/*.mcs.yml` | [Actions_BestPractices.md](./prompts/Actions_BestPractices.md) |
| Settings | `settings.mcs.yml` | [BotSettings_BestPractices.md](./prompts/BotSettings_BestPractices.md) |
| Conversation Mgmt | `StartOver.mcs.yml`, `ResetConversation.mcs.yml` | [ConversationManagement_BestPractices.md](./prompts/ConversationManagement_BestPractices.md) |
| Agent Config | `agent.mcs.yml` | [AgentConfiguration_BestPractices.md](./prompts/AgentConfiguration_BestPractices.md) |
| Entities | `entities/*.mcs.yml` | [Entities_BestPractices.md](./prompts/Entities_BestPractices.md) |

Each prompt is standalone and can be used independently for focused component optimization and improvement.

copilot-studio-agent-optimizer

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README