Changelog

Changelog

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

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

Unreleased

Planned

  • A2A communication hardening: ResponseTracker, HTTP transport SPI, Inbox server, metrics, remote A2A support.

  • Unified AgentIdentity with registry-verified connect flows (DID-JWT/challenge) and route grants.

  • Message v2 (payload, correlation_id, conversation_id), reply()/ignore() helpers.

  • Hub-owned queue service and agent pull workers.

  • Per-hub MCP communication server and hub-owned auth.

  • Protocol adapters: Google A2A, IBM ACP, Google AP2 (outbound adapters and inbound gateways).

  • Composite runtime (single ASGI) and improved CLI UX (serve runtime, human chat/send, register/unregister).

0.4.0 - 2026-07-02

Added

  • New AgentProfile, Skill classes in types.py for comprehensive agent configuration

  • Enhanced capability discovery service using Qdrant vector database

  • Improved agent search tool with filtering options and customizable result detail levels

  • New folder structure for capability discovery implementation

  • Dedicated folder for collaboration tools with improved organization

  • MCP (Model Context Protocol) server (agentconnect.mcp) implementation for agent discovery and search tool

  • Registry API server (agentconnect.servers) with FastAPI-based REST endpoints for agent operations

  • Registry API client (agentconnect.clients) with async HTTP interface and retry logic

  • Centralized search module (agentconnect.core.registry.search) with unified schemas and utilities

  • Centralized SDK configuration system (agentconnect.config) with a single global settings object, YAML-based overrides, and helper utilities to generate, inspect, and validate configuration files

  • New configuration CLI: agentconnect config {init, show, validate}

  • Standalone server configuration via environment variables with production-friendly defaults and vector store options

  • Comprehensive test suite for clients, servers, config, and MCP integration

Changed

  • refactor(cli): new CLI with config, serve, mcp, ping, and doctor commands. Legacy agentconnect/cli.py removed; CLI now lives under agentconnect/cli/ and console script remains agentconnect.

  • Simplified constructor for BaseAgent with focus on profile-based configuration

  • Updated AIAgent, HumanAgent, TelegramAgent constructor to support both profile-based and parameter-based initialization

  • Enhanced AgentRegistration class with additional fields for advanced searches

  • Refactored examples to use the new profile-based agent configuration

  • Reorganized package dependencies in pyproject.toml

  • AgentRegistration converted from dataclass to Pydantic BaseModel

  • CommunicationHub to support both local and remote registry clients

  • Agent search tool to use centralized search schemas

  • Project dependencies updated with MCP protocol and FastAPI support

  • Adopted centralized settings across clients, agents, registry, communication hub, MCP, and utils

  • Unified vector store configuration with in-memory, local-file, and remote modes

  • Environment example files reorganized for clarity: root example.env for SDK-level variables; new demos/.env.example for demo-local secrets

Deprecated

  • Logging policy finalized: library uses NullHandler only; no handlers are added. MCP logs via Context; registry logs use uvicorn logger; CLI/examples control logging level per run. Experimental logging helpers removed.

Removed

  • SDK logging helpers and related config. Library no longer adds handlers or changes third-party loggers.

  • Legacy core configuration module, replaced by centralized settings

  • Temporary payment constants, consolidated into configuration

Fixed

  • Corrected release dates in CHANGELOG.md from 2024 to 2025

  • Agent unregistration to properly call registry.unregister() method

  • Registry index management for capabilities and organizations

  • Agent registration update process to handle all fields correctly

  • Updated Google provider and types to reflect Gemini model changes

Known Limitations

  • Identity verification and registration flows are not yet unified; registry-verified connect (DID-JWT/challenge) and route grants will land post-0.4.0.

  • Message correlation remains metadata-based and will move to a first-class correlation_id in Message v2.

  • Current hub.register(agent) API will be replaced by agent.register(hub) with hub-owned queues and agent pull workers.

  • Registry registration currently accepts full identity payloads; future versions will only accept a public view plus proof and will return a short-lived route grant.

  • Only Discovery MCP is available in 0.4.0 (search/browse). Communication MCP (send/check) arrives in a later 0.x release.

  • Concurrency/backpressure in the hub is basic; ResponseTracker and queue backends arrive later.

  • Only local A2A is supported; remote A2A transport and inbox arrive later.

  • agentconnect.yaml exists but will be hardened and expanded in upcoming releases.

0.3.0 - 2025-05-02

Added

  • Agent payment capabilities using Coinbase Developer Platform (CDP) SDK and AgentKit.

  • Wallet persistence for agents (agentconnect.utils.wallet_manager).

  • CDP environment validation and payment readiness checks (agentconnect.utils.payment_helper).

  • Payment-related dependencies (cdp-sdk, coinbase-agentkit, coinbase-agentkit-langchain) to pyproject.toml.

  • Payment address (payment_address) field to AgentMetadata and AgentRegistration.

  • Payment capability template (PAYMENT_CAPABILITY_TEMPLATE) for agent prompts.

  • Autonomous workflow demo (examples/autonomous_workflow) showcasing inter-agent payments.

  • Standalone chat method in AIAgent for using agents without registration to a hub.

  • Response callbacks in HumanAgent for integration with external systems.

  • Asynchronous collaboration request status checking tool for handling delayed agent responses.

  • Enhanced reasoning step visualization in ToolTracerCallbackHandler for all LLM providers.

  • Comprehensive end-to-end developer guides in documentation.

  • Improved model support for OpenAI and Google AI.

Changed

  • BaseAgent and AIAgent to initialize wallet provider and AgentKit conditionally based on enable_payments flag.

  • AIAgent workflow initialization to include AgentKit tools when payments are enabled.

  • Agent prompts (CORE_DECISION_LOGIC, ReAct prompts) updated to include payment instructions and context.

  • CommunicationHub registration updated to include payment_address.

  • ToolTracerCallbackHandler enhanced to provide specific tracing for payment tool actions.

  • Enhanced HumanAgent to function as an independent agent in the network with run() method.

  • Added stop() method to BaseAgent for proper cleanup and resource management.

  • CommunicationHub improved to handle late responses and request timeouts gracefully.

  • Simplified and enhanced prompt templates for more customizable agents.

  • Updated all examples to use the stop() method for better cleanup.

  • Refactored postprocessing logic in agent_prompts.py for better error handling.

Deprecated

  • examples/run_example.py in favor of using the official CLI tool (agentconnect command)

Fixed

  • Suppressed unnecessary warnings in capability discovery module

  • Improved error handling in agent communication

  • Fixed reasoning step extraction for different LLM providers (OpenAI, Anthropic, Google)

  • Addressed collaboration request timeouts with new polling mechanism

Security

  • Enhanced validation for inter-agent messages

  • Implemented secure API key management

  • Added rate limiting for API calls

  • Set up environment variable handling

  • Added input validation for all API endpoints

0.2.0 - 2025-04-01

Added

  • Research dependency group (poetry install --with research) for enhanced research capabilities

  • Modular multi-agent system with separate specialized agent implementations

  • Enhanced research assistant example with improved web search capabilities

  • Vector embedding support in the agent registry for semantic search capabilities

  • ‘max_turns’ configuration parameter for AIAgent class to limit conversation length

Changed

  • Refactored Telegram assistant into modular multi-agent system with independent components

  • Updated CLI to use the new modular multi-agent system

  • Updated documentation and examples to reflect new architecture

  • Enhanced registry with vector embeddings for more accurate agent capability matching

  • Delegated tool implementations to specialized custom_tools modules for better separation of concerns

  • Optimized prompt templates to reduce redundancy and improve maintainability

  • Updated Telegram agent implementation with improved configuration options

  • Updated project dependencies to latest versions for improved performance and security

Deprecated

  • examples/run_example.py in favor of using the official CLI tool (agentconnect command)

Fixed

  • Suppressed unnecessary warnings in capability discovery module

  • Improved error handling in agent communication

Security

  • Enhanced validation for inter-agent messages

0.1.0 - 2025-03-13

Added

  • Core agent management system with multiple AI provider support:

    • OpenAI integration

    • Anthropic integration

    • Groq integration

    • Google AI integration

  • Multi-agent communication framework:

    • Asynchronous message passing

    • Structured conversation handling

    • Agent state management

  • Example applications:

    • Basic chat with AI assistant

    • Multi-agent e-commerce analysis

    • Research assistant with multiple agents

    • Data analysis and visualization assistant

  • Documentation:

    • Installation and setup guides

    • API reference

    • Usage examples

    • Contributing guidelines

  • Development tooling:

    • Poetry-based dependency management

    • Automated testing with pytest

    • GitHub Actions workflows

    • Code quality tools (black, flake8, pylint)

  • Security features:

    • Environment variable management

    • API key handling

    • Rate limiting support

Security

  • Implemented secure API key management

  • Added rate limiting for API calls

  • Set up environment variable handling

  • Added input validation for all API endpoints