You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: learn-pr/wwl-data-ai/develop-ai-agents-azure-vs-code/includes/3-explore-development-approaches.md
+1-11Lines changed: 1 addition & 11 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -48,17 +48,7 @@ The VS Code extension is ideal for:
48
48
-**Code-first development** - Edit YAML configurations directly for precise control
49
49
-**Local development** - Work on agent designs offline before deploying to Azure
50
50
51
-
The extension installs directly from the Visual Studio Code Marketplace and connects to your existing Foundry projects.
52
-
53
-
### Extension installation
54
-
55
-
To get started with the Microsoft Foundry for Visual Studio Code extension:
56
-
57
-
1. Open Visual Studio Code
58
-
1. Select Extensions from the left pane, or press <kbd>Ctrl</kbd>+<kbd>Shift</kbd>+<kbd>X</kbd>
59
-
1. Search for and select **Microsoft Foundry**
60
-
1. Select **Install**
61
-
1. Verify the extension installed successfully from the status messages
51
+
The extension installs directly from the Visual Studio Code Marketplace and connects to your existing Foundry projects. Detailed installation and setup steps are covered in the next unit.
Copy file name to clipboardExpand all lines: learn-pr/wwl-data-ai/develop-ai-agents-azure-vs-code/includes/4-build-agent-azure-portal.md
+4-49Lines changed: 4 additions & 49 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -15,56 +15,11 @@ The Foundry portal streamlines agent creation through an intuitive interface:
15
15
16
16
The portal creates your agent and opens the configuration interface where you can refine its behavior and capabilities.
17
17
18
-
## Designing effective agent instructions
18
+
## Configuring agent instructions and properties
19
19
20
-
Agent instructions are the foundation of agent behavior. These instructions define how your agent understands its role, responds to users, and handles various scenarios. Well-crafted instructions lead to consistent, reliable agent behavior.
20
+
Agent instructions are the foundation of agent behavior. In the **Instructions** field, you define how your agent understands its role, responds to users, and handles various scenarios. Clear, specific instructions lead to consistent, reliable agent behavior. You'll configure instructions in more detail when working in Visual Studio Code later in this module.
21
21
22
-
### Writing clear instructions
23
-
24
-
When writing system instructions for your agent:
25
-
26
-
**Be specific and clear** - Define exactly what the agent should do and how it should behave. Vague instructions lead to unpredictable responses.
27
-
28
-
**Provide context** - Explain the agent's role and the environment it operates in. For example: "You're a customer service agent for a healthcare organization. You help patients schedule appointments and answer general questions about services."
29
-
30
-
**Set boundaries** - Clearly define what the agent should and shouldn't do. Include guidelines for handling requests outside its scope.
31
-
32
-
**Include examples** - Show the agent examples of desired interactions when helpful. Examples demonstrate the tone, style, and level of detail you expect.
33
-
34
-
**Define personality** - Establish the tone and style of responses. Should the agent be formal or casual? Technical or accessible?
35
-
36
-
### Instruction example for a customer service agent
37
-
38
-
```text
39
-
You're a helpful customer service agent for Contoso Healthcare. Your role is to assist patients with scheduling appointments, answering questions about our services, and providing general information.
40
-
41
-
When helping patients:
42
-
- Always be polite, professional, and empathetic
43
-
- Confirm appointment details before finalizing bookings
44
-
- If asked about medical advice, politely explain you can't provide medical guidance and suggest consulting with their healthcare provider
45
-
- For complex billing questions, offer to connect them with our billing department
46
-
- Maintain patient privacy and never share personal information
47
-
48
-
Keep responses concise but informative. Use a warm, professional tone that makes patients feel cared for.
49
-
```
50
-
51
-
This instruction provides clear role definition, behavioral guidelines, boundaries, and tone guidance; all essential elements for consistent agent performance.
52
-
53
-
## Configuring agent properties
54
-
55
-
Beyond instructions, agents have several important properties you configure in the portal:
56
-
57
-
### Temperature setting
58
-
59
-
Temperature controls response randomness. Lower values (0.1-0.3) produce consistent, focused responses. Higher values (0.7-1.0) generate more creative, varied outputs. For most business agents, values between 0.3 and 0.7 balance consistency with natural language variation.
60
-
61
-
### Top P setting
62
-
63
-
Top P (nucleus sampling) controls response diversity by limiting the model's vocabulary choices. Values near 1.0 allow more variety, while lower values constrain outputs to more likely options. Most scenarios work well with the default value of 1.0.
64
-
65
-
### Response length limits
66
-
67
-
You can set maximum token limits for responses, preventing overly long outputs and controlling costs. Consider your use case—customer service agents might need shorter, focused responses, while research agents might benefit from longer, detailed answers.
22
+
Beyond instructions, the portal lets you configure model parameters such as **Temperature** (which controls response randomness) and **Top P** (which controls response diversity). These configuration options are explored further in the Visual Studio Code configuration unit.
68
23
69
24
## Testing your agent in the portal
70
25
@@ -82,7 +37,7 @@ Before deployment, you can enhance your agent with basic capabilities:
82
37
83
38
**Bing Search** - Connects your agent to real-time internet information.
84
39
85
-
You add these tools through the Tools section in the agent configuration. Tool capabilities are explored in detail later in this module.
40
+
You add these tools through the Tools section in the agent configuration. Tool capabilities are explored in detail in the *Extend agent capabilities* unit later in this module.
Copy file name to clipboardExpand all lines: learn-pr/wwl-data-ai/develop-ai-agents-azure-vs-code/includes/5-set-up-vs-code.md
-75Lines changed: 0 additions & 75 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,22 +4,6 @@ Setting up Visual Studio Code for AI agent development brings enterprise-grade c
4
4
5
5
The Microsoft Foundry for Visual Studio Code extension provides direct access to Microsoft Foundry Agent Service capabilities. This extension creates an integrated experience for agent development that combines visual design tools with code-based configuration.
6
6
7
-
### Core capabilities
8
-
9
-
The extension offers several key features that streamline agent development:
10
-
11
-
**Agent discovery and management** - Browse, create, and manage AI agents within your Microsoft Foundry projects. View all your agents in the Azure Resources explorer without opening a browser.
12
-
13
-
**Visual agent designer** - Configure agent instructions, tools, and capabilities through an intuitive interface. The designer provides the same power as the Foundry portal while keeping you in your development environment.
14
-
15
-
**Integrated testing** - Test agents in real-time using the built-in playground. Verify behavior and iterate on designs without switching contexts or tools.
16
-
17
-
**Code generation** - Generate sample integration code to connect agents with your applications. The extension creates ready-to-use code snippets that accelerate implementation.
18
-
19
-
**Deployment pipeline** - Deploy agents directly to Microsoft Foundry for production use. Simple commands handle the deployment process, providing feedback about status and results.
20
-
21
-
**Dual configuration modes** - Work with both visual interfaces and direct YAML editing. Switch between modes based on your immediate needs and preferences.
22
-
23
7
24
8
25
9
## Installing and configuring the extension
@@ -87,65 +71,6 @@ With the extension configured and a model deployed, you're ready to create an ag
87
71
88
72
Both views remain synchronized; changes in one automatically update the other. This dual approach lets you work with whichever format feels more natural for your current task.
89
73
90
-
## Understanding the development workflow
91
-
92
-
The Microsoft Foundry extension supports an efficient development workflow:
93
-
94
-
1.**Create or import** agents using the visual designer or existing YAML files
95
-
1.**Configure** agent properties, instructions, and capabilities
96
-
1.**Test** immediately using the integrated playground
97
-
1.**Iterate** on design based on test results
98
-
1.**Deploy** to Azure with a single command
99
-
1.**Generate integration code** for your applications
100
-
1.**Monitor** deployed agents through the Azure Resources view
101
-
102
-
This workflow eliminates context switching between tools, keeping you focused on agent development rather than infrastructure management.
103
-
104
-
## Working with agent YAML files
105
-
106
-
While the visual designer provides an accessible interface, the YAML files give you direct access to agent configuration. Understanding YAML structure helps you make precise changes and version control your agents effectively.
description: Handles customer inquiries and appointment scheduling
117
-
id: ''
118
-
metadata:
119
-
authors:
120
-
- your-name
121
-
tags:
122
-
- customer-service
123
-
- healthcare
124
-
model:
125
-
id: 'gpt-4o-deployment'
126
-
options:
127
-
temperature: 0.7
128
-
top_p: 1
129
-
instructions: |
130
-
You're a helpful customer service agent for a healthcare organization.
131
-
Assist patients with scheduling appointments and answering questions.
132
-
tools: []
133
-
```
134
-
135
-
The YAML format is straightforward, with clear sections for each aspect of your agent's configuration. You can edit this file directly or use the visual designer—both approaches modify the same underlying configuration.
136
-
137
-
### Benefits of YAML configuration
138
-
139
-
Direct YAML editing provides several advantages:
140
-
141
-
- **Version control** - Track changes in Git alongside your application code
142
-
- **Bulk updates** - Make multiple changes simultaneously with confidence
143
-
- **Templates** - Create reusable agent templates for consistent configurations
144
-
- **Code review** - Include agent configurations in your standard code review processes
145
-
- **Automation** - Build scripts that generate or modify agent configurations programmatically
146
-
147
-
The extension validates YAML syntax in real-time, highlighting errors and providing suggestions as you type.
148
-
149
74
## Managing multiple agents
150
75
151
76
As your projects grow, you'll likely manage multiple agents with different purposes. The Microsoft Foundry extension makes this straightforward:
Copy file name to clipboardExpand all lines: learn-pr/wwl-data-ai/develop-ai-agents-azure-vs-code/includes/6-configure-manage-agents.md
+9-122Lines changed: 9 additions & 122 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -73,130 +73,17 @@ tools: []
73
73
74
74
The YAML structure divides naturally into sections: metadata, model configuration, instructions, and tools. This organization makes it easy to locate and modify specific settings.
75
75
76
-
## Writing effective agent instructions
76
+
### Benefits of YAML configuration
77
77
78
-
Agent instructions are the most critical configuration element. They define how your agent understands its role, makes decisions, and responds to users. Well-crafted instructions lead to consistent, reliable agent behavior.
78
+
Direct YAML editing provides several advantages:
79
79
80
-
### Best practices for instruction design
80
+
- **Version control** - Track changes in Git alongside your application code
81
+
- **Bulk updates** - Make multiple changes simultaneously with confidence
82
+
- **Templates** - Create reusable agent templates for consistent configurations
83
+
- **Code review** - Include agent configurations in your standard code review processes
84
+
- **Automation** - Build scripts that generate or modify agent configurations programmatically
81
85
82
-
When writing system instructions:
83
-
84
-
**Be specific and clear** - Define exactly what the agent should do and how it should behave. Ambiguity in instructions leads to unpredictable behavior. Instead of "Help users with appointments," write "Assist healthcare staff in scheduling patient appointments by finding available time slots and confirming booking details."
85
-
86
-
**Provide context** - Explain the agent's role and operating environment. Context helps the agent understand the broader situation and make better decisions. For example: "You work in a busy healthcare facility where staff need quick, accurate appointment information."
87
-
88
-
**Set boundaries** - Clearly define what the agent should and shouldn't do. Boundaries prevent the agent from attempting tasks outside its scope or capabilities. Include statements like "Never access patient medical records" or "Don't provide medical advice."
89
-
90
-
**Include examples** - Show the agent examples of desired interactions when helpful. Examples demonstrate the level of detail, tone, and format you expect in responses.
91
-
92
-
**Define personality** - Establish the tone and style of responses. Should the agent be formal or casual? Technical or accessible? Professional but warm, or strictly business-focused?
93
-
94
-
**Specify escalation paths** - Tell the agent what to do when it encounters situations it can't handle. For example: "If staff ask about complex billing disputes, acknowledge the question and suggest they contact the billing department directly."
95
-
96
-
### Instruction examples for different scenarios
97
-
98
-
**Technical support agent:**
99
-
100
-
```text
101
-
You're a technical support agent helping users troubleshoot software issues.
102
-
103
-
Your approach:
104
-
- Start by asking clarifying questions to understand the problem
105
-
- Guide users through systematic troubleshooting steps
106
-
- Explain technical concepts in simple terms
107
-
- Document solutions that work for future reference
108
-
109
-
When users report issues:
110
-
- Ask about error messages, recent changes, and affected features
111
-
- Suggest basic troubleshooting before advanced solutions
112
-
- If the issue requires escalation, gather all relevant details first
113
-
114
-
Keep responses clear and actionable. Use a patient, supportive tone; users may be frustrated when technical issues disrupt their work.
115
-
```
116
-
117
-
**Sales support agent:**
118
-
119
-
```text
120
-
You're a sales support agent helping the sales team research leads and prepare for meetings.
121
-
122
-
Your tasks:
123
-
- Research company information when given a company name or domain
124
-
- Summarize key facts relevant to potential sales opportunities
125
-
- Identify recent news or events that might be conversation starters
126
-
- Suggest relevant talking points based on the company's industry and size
127
-
128
-
When researching companies:
129
-
- Focus on publicly available information
130
-
- Highlight information relevant to our product offerings
131
-
- Note any recent executive changes, funding rounds, or expansion news
132
-
- Keep summaries concise, sales teams need quick insights
133
-
134
-
Maintain a professional tone focused on actionable intelligence that helps close deals.
135
-
```
136
-
137
-
## Deploying agents from Visual Studio Code
138
-
139
-
Once you're satisfied with your agent's configuration, deploy it to Microsoft Foundry to make it available for production use.
140
-
141
-
### Deployment process
142
-
143
-
To deploy your agent:
144
-
145
-
1. Open your agent in the Agent Designer
146
-
1. Verify all configuration is correct
147
-
1. Select **Create on Microsoft Foundry** (button in the bottom-left of the Designer)
148
-
1. Wait for deployment to complete — status updates appear in VS Code notifications
149
-
1. After successful deployment, refresh the Azure Resources view
150
-
1. Your deployed agent appears under the Agents subsection
151
-
152
-
The deployment process uploads your agent configuration to Microsoft Foundry, making it accessible through APIs and other applications.
153
-
154
-
### Updating deployed agents
155
-
156
-
When you need to modify a deployed agent:
157
-
158
-
1. Select the deployed agent in the Azure Resources view
159
-
1. The Agent Designer opens showing current configuration
160
-
1. Make your changes in either the Designer or YAML file
161
-
1. Select **Update on Microsoft Foundry** to deploy the changes
162
-
1. Wait for the update to complete
163
-
164
-
Updates replace the existing configuration with your new version. The agent ID remains the same, so existing integrations continue working with the updated behavior.
165
-
166
-
## Managing deployed agents
167
-
168
-
After deployment, the extension provides several management capabilities:
169
-
170
-
### Viewing agent details
171
-
172
-
Select a deployed agent to see its configuration, deployment status, and available actions. The Agent Preferences page shows:
173
-
- Current configuration and instructions
174
-
- Model deployment being used
175
-
- Tools and capabilities enabled
176
-
- Deployment timestamp and status
177
-
178
-
### Generating integration code
179
-
180
-
The extension can generate sample code for integrating agents into applications:
181
-
182
-
1. Select a deployed agent in the Azure Resources view
183
-
1. Select **Open Code File** from the available actions
184
-
1. Choose your programming language (Python, C#, JavaScript, etc.)
185
-
1. The extension generates code showing how to connect to and use the agent
186
-
187
-
This generated code includes authentication, connection setup, and example interactions, accelerating integration into your applications.
188
-
189
-
### Testing deployed agents
190
-
191
-
The integrated playground lets you test deployed agents:
192
-
193
-
1. Select a deployed agent
194
-
1. Select **Open Playground**
195
-
1. The playground opens in VS Code
196
-
1. Start a conversation to test the agent's behavior
197
-
1. Verify responses match your expectations
198
-
199
-
The playground uses the deployed agent configuration, ensuring your tests reflect production behavior.
86
+
The extension validates YAML syntax in real-time, highlighting errors and providing suggestions as you type.
200
87
201
88
## Best practices for agent configuration
202
89
@@ -214,4 +101,4 @@ As you build more complex agents, these practices help maintain quality and reli
214
101
215
102
**Keep instructions focused** - Each agent should have a clear, specific purpose. Agents trying to do too many things perform inconsistently.
216
103
217
-
Configuring and managing agents in Visual Studio Code provides powerful capabilities for creating sophisticated automation. The combination of visual design tools, direct YAML editing, and integrated testing enables rapid development while maintaining the precision needed for production deployments. With these configuration techniques, you can build agents that reliably meet your specific requirements.
104
+
Configuring agents in Visual Studio Code provides powerful capabilities for creating sophisticated automation. The combination of visual design tools and direct YAML editingenables rapid development while maintaining the precision needed for production deployments.
0 commit comments