Skip to content

Latest commit

 

History

History
326 lines (257 loc) · 10.1 KB

CHANGELOG.md

File metadata and controls

326 lines (257 loc) · 10.1 KB

Changelog

All notable changes to this project will be documented in this file.

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

[1.0.0] - 2025-11-21

Major refactor with breaking changes. Complete provider rewrite. New modular architecture.

Requirements: Ruby 3.1+, Rails 7.0+/8.0+/8.1+

What's Changed

  • Major Framework Refactor: ActiveAgent v1.0.0 by @sirwolfgang in #259
  • Add API gem version testing and fix Anthropic 1.14.0 compatibility by @sirwolfgang in #265
  • Fix version compatiblity issue for vitepress by @sirwolfgang in #266
  • Add missing API Keys by @sirwolfgang in #267
  • Fix website links by @sirwolfgang in #268
  • chore: remove standard from dev dependencies by @okuramasafumi in #272
  • Add thread safety tests by @sirwolfgang in #275
  • Refactor: Leverage Native Gem Types Across All Providers by @sirwolfgang in #271
  • Improved Usage Tracking by @sirwolfgang in #274

New Contributors

  • @okuramasafumi made their first contribution in #272

Full Changelog: https://github.com/activeagents/activeagent/compare/v0.6.3...v1.0.0

Added

Universal Tools Format

# Single format works across all providers (Anthropic, OpenAI, OpenRouter, Ollama, Mock)
tools: [{
  name: "get_weather",
  description: "Get current weather",
  parameters: {
    type: "object",
    properties: {
      location: { type: "string", description: "City and state" }
    },
    required: ["location"]
  }
}]

# Tool choice normalization
tool_choice: "auto"                   # Let model decide
tool_choice: "required"               # Force tool use
tool_choice: { name: "get_weather" }  # Force specific tool

Automatic conversion to provider-specific formats. Old formats still work (backward compatible).

Model Context Protocol (MCP) Support

# Universal MCP format works across providers (Anthropic, OpenAI)
class MyAgent < ActiveAgent::Base
  generate_with :anthropic, model: "claude-haiku-4-5"

  def research
    prompt(
      message: "Research AI developments",
      mcps: [{
        name: "github",
        url: "https://api.githubcopilot.com/mcp/",
        authorization: ENV["GITHUB_MCP_TOKEN"]
      }]
    )
  end
end
  • Common format: {name: "server", url: "https://...", authorization: "token"}
  • Auto-converts to provider native formats
  • Anthropic: Beta API support, up to 20 servers per request
  • OpenAI: Responses API with pre-built connectors (Dropbox, Google Drive, etc.)
  • Backwards compatible: accepts both mcps and mcp_servers parameters
  • Comprehensive documentation with tested examples
  • Full VCR test coverage with real MCP endpoints

Changed

  • Shared ToolChoiceClearing concern eliminates duplication across providers

Breaking Changes

1. Update Provider Gems

# Gemfile - Remove unofficial gems
gem "ruby-openai"
gem "ruby-anthropic"

# Add official provider SDKs
gem "openai"      # Official OpenAI SDK
gem "anthropic"   # Official Anthropic SDK

Run bundle install after updating.

2. Update Base Class

# Before
class MyAgent < ActiveAgent::ActionPrompt::Base
end

# After
class MyAgent < ActiveAgent::Base
end

3. Configure Providers

# Before - options wrapped in options key
class MyAgent < ActiveAgent::Base
  def chat
    prompt(message: "Hello", options: { temperature: 0.7 })
  end
end

# After - options passed directly (at class or call level)
class MyAgent < ActiveAgent::Base
  generate_with :openai, model: "gpt-4o-mini", temperature: 0.7

  def chat
    prompt("Hello")  # Uses class-level config
  end

  def chat_creative
    prompt("Hello", temperature: 1.0)  # Override per-call
  end
end

4. Update Custom Providers (if any)

# Before
module ActiveAgent::GenerationProvider
  class CustomProvider < Base
  end
end

# After
module ActiveAgent::Providers
  class CustomProvider < BaseProvider
  end
end

5. Update Generator Commands

# Before
rails g active_agent MyAgent action

# After
rails g active_agent:agent MyAgent action

6. Remove Framework Retry Config

# Remove from config/initializers/activeagent.rb
ActiveAgent.configure do |config|
  config.retries = true
  config.retries_count = 5
end

# Use provider-specific settings in config/active_agent.yml
openai:
  service: "OpenAI"
  max_retries: 5
  timeout: 600.0

Template paths:

  • app/views/agents/{agent}/instructions.md (no .erb extension by default for instructions)
  • app/views/agents/{agent}/{action}.md.erb

Added

Mock Provider for Testing

class MyAgent < ActiveAgent::Base
  generate_with :mock
end

response = MyAgent.prompt("Test").generate_now
# Returns predictable responses without API calls

Mixed Provider Support

class MyAgent < ActiveAgent::Base
  generate_with :openai, model: "gpt-4o-mini"
  embed_with :anthropic, model: "claude-3-5-sonnet-20241022"
end

Prompt Previews

preview = MyAgent.prompt("Hello").prompt_preview
# Shows instructions, messages, tools before execution

Callback Lifecycle

  • before_generation, after_generation, around_generation
  • before_prompt, after_prompt, around_prompt
  • before_embed, after_embed, around_embed
  • on_stream_open, on_stream, on_stream_close
  • Rails-style callback control: prepend_*, skip_*, append_*

Multi-Input Embeddings

response = MyAgent.embed(inputs: ["Text 1", "Text 2"]).embed_now
vectors = response.data.map { |d| d[:embedding] }

Normalized Usage Statistics

response = MyAgent.prompt("Hello").generate_now

# Works across all providers
response.usage.input_tokens
response.usage.output_tokens
response.usage.total_tokens

# Provider-specific fields when available
response.usage.cached_tokens      # OpenAI, Anthropic
response.usage.reasoning_tokens   # OpenAI o1 models
response.usage.service_tier       # Anthropic

Enhanced Instrumentation for APM Integration

  • Unified event structure: prompt.active_agent and embed.active_agent (top-level) plus prompt.provider.active_agent and embed.provider.active_agent (per-API-call)
  • Event payloads include comprehensive data for monitoring tools (New Relic, DataDog, etc.):
    • Request parameters: model, temperature, max_tokens, top_p, stream, message_count, has_tools
    • Usage data: input_tokens, output_tokens, total_tokens, cached_tokens, reasoning_tokens, audio_tokens, cache_creation_tokens (critical for cost tracking)
    • Response metadata: finish_reason, response_model, response_id, embedding_count
  • Top-level events report cumulative usage across all API calls in multi-turn conversations
  • Provider-level events report per-call usage for granular tracking

Multi-Turn Usage Tracking

  • response.usage now returns cumulative token counts across all API calls during tool calling
  • New response.usages array contains individual usage objects from each API call
  • Usage objects support addition: usage1 + usage2 for combining statistics

Provider Enhancements

  • OpenAI Responses API: api: :responses or api: :chat
  • Anthropic JSON object mode with automatic extraction
  • OpenRouter: quantization, provider preferences, web search
  • Flexible naming: :openai or :open_ai, :openrouter or :open_router

Rails 8.1 Support

Comprehensive Documentation

  • VitePress site at docs.activeagents.ai
  • All examples tested and validated

Changed

Provider Architecture

  • Unified BaseProvider interface across all providers
  • Retry logic moved to provider SDKs (automatic exponential backoff)
  • Migrated to official SDKs: openai gem and anthropic gem
  • Type-safe options with per-provider definitions

Configuration

  • Options configurable at class level, instance level, or per-call
  • Simplified parameter handling pattern

Requirements

  • Ruby 3.1+ (previously 3.0+)

Testing

  • Reorganized by feature and provider integration
  • All documentation examples validated

Fixed

Providers

  • OpenAI streaming with functions/tools
  • Ollama streaming support
  • Anthropic tool choice modes (any and tool)
  • OpenRouter model fallback and parameter naming
  • Provider gem loading errors

Framework

  • Streaming lifecycle with function/tool calls
  • Multi-tool and multi-turn conversation handling
  • Options mutation during generation
  • Template rendering without blocks
  • Schema generator key symbolization
  • Rails 8.0 and 8.1 compatibility
  • Usage extraction across OpenAI/Anthropic response formats

Removed

Namespaces

  • ActiveAgent::ActionPrompt → use ActiveAgent::Base
  • ActiveAgent::GenerationProvider → use ActiveAgent::Providers

Configuration

  • ActiveAgent.configuration.retries → use provider max_retries
  • ActiveAgent.configuration.retries_count → use provider max_retries
  • ActiveAgent.configuration.retries_on → handled by provider SDKs

Modules

  • ActiveAgent::QueuedGeneration → Queueing concern
  • ActiveAgent::Rescuable → Rescue concern
  • ActiveAgent::Sanitizers → moved to concerns
  • ActiveAgent::PromptHelper → moved to concerns

[0.3.2] - 2025-04-15

Added

  • CI configuration for stable GitHub releases moving forward.
  • Test coverage for core features: ActionPrompt rendering, tool calls, and embeddings.
  • Enhance streaming to support tool calls during stream. Previously, streaming mode blocked tool call execution.
  • Fix layout rendering bug when no block is passed and views now render correctly without requiring a block.

Removed

  • Generation Provider module and Action Prompt READMEs have been removed, but will be updated along with the main README in the next release.