Habilidades Azure.AI.Agents.Persistent (.NET)
📦

Azure.AI.Agents.Persistent (.NET)

Seguro

Build persistent AI agents with Azure AI SDK for .NET

Developers need to integrate persistent AI agents into .NET applications with proper thread management and streaming. This skill provides complete SDK patterns for agent CRUD, conversation flows, function calling, and file search using Azure AI Agents.

Soporta: Claude Codex Code(CC)
🥉 72 Bronce
1

Descargar el ZIP de la skill

2

Subir en Claude

Ve a Configuración → Capacidades → Skills → Subir skill

3

Activa y empieza a usar

Pruébalo

Usando "Azure.AI.Agents.Persistent (.NET)". Create a math tutor agent with code interpreter enabled

Resultado esperado:

Agent created successfully with ID 'asst_abc123'. The agent is configured with gpt-4o-mini model and CodeInterpreterTool. Instructions: 'You are a personal math tutor. Write and run code to answer math questions.' Ready to create threads and process user messages.

Usando "Azure.AI.Agents.Persistent (.NET)". Stream a response from the agent for user question about solving 3x + 11 = 14

Resultado esperado:

Streaming output: 'To solve 3x + 11 = 14, I need to isolate x. First, subtract 11 from both sides: 3x = 3. Then divide by 3: x = 1. Let me verify this with code.' Run completed with final answer showing x equals 1.

Auditoría de seguridad

Seguro
v1 • 2/24/2026

Documentation-only skill with no executable code. Static analysis found 0 security issues with risk score 0/100. The SKILL.md file contains legitimate Azure SDK usage examples for agent creation, thread management, streaming, and function calling. No security concerns identified.

0
Archivos escaneados
0
Líneas analizadas
0
hallazgos
1
Auditorías totales
No se encontraron problemas de seguridad
Auditado por: claude

Puntuación de calidad

38
Arquitectura
100
Mantenibilidad
87
Contenido
50
Comunidad
100
Seguridad
74
Cumplimiento de la especificación

Lo que puedes crear

Build a Math Tutor Agent

Create a persistent agent that solves math problems using code interpreter. Users submit questions via thread messages, and the agent executes Python code to compute answers with step-by-step explanations.

Document Search Assistant

Implement an agent with vector store file search to answer questions from uploaded documentation. Upload files, create vector stores, and query through natural language conversations.

Weather Bot with Function Calling

Build an agent that calls external weather APIs through function definitions. Define tool schemas, handle RequiresAction responses, and return formatted weather data to users.

Prueba estos prompts

Create a Basic Agent 
Create a persistent AI agent using Azure.AI.Agents.Persistent SDK. The agent should use gpt-4o-mini model with instructions to act as a helpful assistant. Show authentication with DefaultAzureCredential and basic agent creation code.
Implement Streaming Responses 
Show me how to stream agent responses in real-time using CreateRunStreamingAsync. Include handling for MessageContentUpdate and run status events like RunCreated and RunCompleted.
Add Function Calling to Agent 
Create an agent with a custom function tool for getting weather data. Define the FunctionToolDefinition with JSON schema parameters, handle RequiresAction status, and submit tool outputs back to the run.
Configure File Search with Vector Store 
Set up file search capabilities for an agent. Upload a document file, create a vector store with the file ID, configure FileSearchToolResource, and create an agent that can search documents to answer questions.

Mejores prácticas

  • Always dispose clients using 'using' statements or explicit disposal to release resources
  • Poll agent runs with 500ms delays between status checks to avoid rate limiting
  • Delete threads, agents, and vector stores when finished to clean up resources

Evitar

  • Do not skip error handling for RequestFailedException when calling agent APIs
  • Avoid hardcoding credentials or endpoint URLs directly in application code
  • Avoid tight polling loops without delays which can trigger rate limits

Preguntas frecuentes

What is the difference between PersistentAgentsClient and Azure.AI.Projects?
PersistentAgentsClient provides low-level direct access to agent operations. Azure.AI.Projects offers a higher-level abstraction that wraps the persistent agents functionality with additional project management features.
How do I handle function calling in a streaming run?
Function calling requires polling pattern, not streaming. When run status is RequiresAction, extract the function call details, execute your function, submit tool outputs, then continue polling until completion.
What authentication methods are supported?
The SDK supports DefaultAzureCredential which includes managed identity, Azure CLI credentials, Visual Studio credentials, and service principal authentication via environment variables.
How long do threads and messages persist?
Threads and messages persist indefinitely until explicitly deleted. Store thread IDs to resume conversations later. Consider implementing cleanup logic for unused threads to manage costs.
Can I use multiple tools with a single agent?
Yes, you can configure multiple tools when creating an agent. Pass an array containing CodeInterpreterToolDefinition, FileSearchToolDefinition, FunctionToolDefinition, and other tool types as needed.
What are the rate limits for agent operations?
Rate limits depend on your Azure AI resource tier. Implement exponential backoff for RequestFailedException with status 429. Default polling interval of 500ms is generally safe for most scenarios.

Detalles del desarrollador

Autor

sickn33

Licencia

MIT

Repositorio

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

Ref.

main

Estructura de archivos

📄 SKILL.md