Skills Azure AI Agents Persistent SDK for Java
📦

Azure AI Agents Persistent SDK for Java

Safe

Create persistent AI agents with Azure SDK

Build persistent AI agents that maintain state across sessions. This SDK provides low-level control over agent threads, messages, and runs in Java applications.

Supports: Claude Codex Code(CC)
📊 69 Adequate
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "Azure AI Agents Persistent SDK for Java". Create a math tutor agent that helps students with algebra

Expected outcome:

  • Agent 'Math Tutor' created successfully with ID: agent_abc123
  • Thread created with ID: thread_xyz789
  • Message added to thread
  • Run initiated and completed
  • Assistant: To solve 2x + 5 = 15, first subtract 5 from both sides...

Using "Azure AI Agents Persistent SDK for Java". Execute code to solve a quadratic equation

Expected outcome:

  • Run status: RequiresAction
  • Tool call: code_interpreter
  • Code executed: numpy.roots([1, -5, 6])
  • Assistant: The solutions are x = 2 and x = 3

Security Audit

Safe
v1 • 2/24/2026

Static analysis scanned 0 files with 0 lines. Risk score: 0/100. This skill contains only documentation (SKILL.md) describing the Azure AI Agents Persistent SDK for Java. No executable code, scripts, or security-sensitive patterns detected. The documentation describes standard Azure authentication patterns using DefaultAzureCredential which is the recommended secure approach.

0
Files scanned
0
Lines analyzed
0
findings
1
Total audits
No security issues found
Audited by: claude

Quality Score

38
Architecture
100
Maintainability
87
Content
31
Community
100
Security
74
Spec Compliance

What You Can Build

Customer Support Agent Integration

Embed a persistent support agent in your Java backend that maintains conversation history across customer sessions.

Educational Tutoring System

Build a personalized tutoring agent that tracks student progress and adapts teaching style over multiple sessions.

Enterprise Assistant Workflow

Create reusable agent instances for common enterprise tasks like document processing or data analysis.

Try These Prompts

Basic Agent Setup 
Help me create a basic Azure AI persistent agent using Java. Show me how to set up authentication with DefaultAzureCredential, create an agent with custom instructions, and send a simple message.
Multi-turn Conversation 
Show me how to build a multi-turn conversation flow with Azure AI persistent agents. Include thread creation, adding messages, running the agent, polling for completion, and retrieving all messages from the thread.
Agent with Custom Tools 
I need to create an Azure AI agent that uses custom tools. Show me how to define tools, attach them to an agent, handle the RequiresAction status, and submit tool outputs back to the run.
Production Agent Workflow 
Create a complete production-ready example of Azure AI persistent agents in Java. Include proper error handling for all run statuses, async client usage for high throughput, resource cleanup in finally blocks, and environment-based configuration.

Best Practices

  • Use DefaultAzureCredential for secure production authentication instead of hardcoded keys
  • Implement proper polling delays (500ms recommended) between run status checks to avoid rate limiting
  • Always clean up resources by deleting threads and agents when no longer needed

Avoid

  • Skipping polling between run status checks which causes missed state transitions
  • Hardcoding credentials instead of using environment variables and Azure managed identities
  • Not handling all run statuses (RequiresAction, Failed, Cancelled) leading to stuck workflows

Frequently Asked Questions

What is the Azure AI Agents Persistent SDK?
The Azure AI Agents Persistent SDK for Java provides a low-level API for creating and managing persistent AI agents. Unlike stateless APIs, persistent agents maintain threads and conversation history across multiple sessions, enabling complex multi-turn interactions.
How do I authenticate with Azure AI Agents?
Use DefaultAzureCredential which automatically handles authentication through environment variables, managed identities, or Azure CLI credentials. Set the PROJECT_ENDPOINT environment variable to your Azure AI project endpoint URL.
What is the difference between an agent, thread, and run?
An agent is the configured AI instance with specific instructions and tools. A thread is the conversation container that holds messages. A run is the execution of the agent against a thread, processing messages and generating responses.
Do persistent agents survive application restarts?
Yes, agents and threads are stored server-side in Azure. You can retrieve existing agents and threads by their IDs across application sessions, enabling true persistence of conversations and agent configurations.
How do I handle tool calls with agents?
When an agent requires tool execution, the run status becomes RequiresAction. You must submit the tool outputs using submitToolOutputsToRun, then continue polling the run status until completion.
Is there an async version of the SDK?
Yes, use PersistentAgentsAsyncClient instead of PersistentAgentsClient for non-blocking operations. The async client is recommended for high-throughput scenarios and better resource utilization.

Developer Details

Author

sickn33

License

MIT

Repository

https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/azure-ai-agents-persistent-java

Ref

main

File structure

📄 SKILL.md