Skip to content

CHANGELOG.md

Latest commit

 

History

History
92 lines (70 loc) · 3.05 KB

CHANGELOG.md

File metadata and controls

92 lines (70 loc) · 3.05 KB

Changelog

All notable changes to WorkflowCore.AI.AzureFoundry will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.0.0-beta.1] - 2026-01-27

Added

  • ChatCompletion Step - Invoke Azure AI models with conversation history support

    • Configurable system prompts, temperature, and max tokens
    • Automatic conversation history management
    • Token usage tracking for cost monitoring

  • AgentLoop Step - Agentic workflows with automatic tool execution

    • LLM-driven tool selection and invocation
    • Configurable iteration limits to prevent runaway loops
    • Support for both automatic and manual tool execution modes
    • Tool result tracking and debugging

  • ExecuteTool Step - Manual tool execution for fine-grained control

    • Direct tool invocation by name with JSON arguments
    • Error handling with success/failure results

  • GenerateEmbedding Step - Vector embedding generation

    • Support for Azure AI embedding models
    • Configurable model selection
    • Token usage tracking

  • VectorSearch Step - Semantic search with Azure AI Search

    • Vector similarity search
    • OData filter support
    • Configurable result count (TopK)

  • HumanReview Step - Human-in-the-loop approval workflows

    • Pause workflow for human review
    • Support for approve, reject, and modify actions
    • Configurable reviewer assignment and prompts

  • Tool Framework

    • IAgentTool interface for custom tool implementations
    • IToolRegistry for tool registration and discovery
    • JSON Schema parameter definitions for tool calling
    • ToolResult with success/failure states

  • Conversation History Management

    • IConversationStore abstraction for pluggable storage
    • InMemoryConversationStore default implementation
    • Automatic thread management per workflow execution
    • ConversationMessage and ConversationThread models

  • Azure AI Foundry Integration

    • Support for Azure AI Foundry (services.ai.azure.com) endpoints
    • API key and Azure AD authentication
    • Configurable default models and parameters
    • Azure AI Search integration for RAG scenarios

  • Fluent Builder API

    • ChatCompletion() extension method
    • AgentLoop() extension method
    • GenerateEmbedding() extension method
    • VectorSearch() extension method
    • HumanReview() extension method

Dependencies

  • Azure.AI.Inference 1.0.0-beta.5
  • Azure.AI.Projects 1.0.0-beta.2
  • Azure.Identity 1.13.0
  • Azure.Search.Documents 11.6.0

Notes

  • This is a beta release - APIs may change before 1.0.0 stable
  • Requires .NET Standard 2.0 or higher
  • Compatible with WorkflowCore 3.x

[Unreleased]

Planned Features

  • Streaming response support for real-time output
  • Structured output with JSON schema validation
  • Vision/multimodal input support
  • OpenTelemetry tracing integration
  • Rate limiting and retry configuration
  • Batch embedding generation
  • More conversation store implementations (Redis, SQL, CosmosDB)