v0.7.0

Release v0.7.0 includes:\n- Refactored tests to use standard pytest assertions\n- Added CostMetric and TokenUsageMetric\n- Added ToolEmulator for agent testing\n- Enhanced Dashboard with Comparison Report\n- Bumped version to 0.7.0

v0.6.0 - Open Source Readiness

AgentUnit v0.6.0 - Open Source Readiness Release

This release focuses on making AgentUnit ready for open source collaboration, with comprehensive documentation, developer tooling, and community infrastructure.

Highlights

Documentation & Guides

• CONTRIBUTING.md with comprehensive contribution guidelines
• SECURITY.md with responsible vulnerability disclosure process
• Adapter implementation guide for all supported frameworks
• Complete metrics catalog reference
• OpenTelemetry telemetry configuration guide
• Comparison documentation vs Ragas, DEEPEVAL, and other frameworks
• Sample benchmark suite with RAG evaluation dataset

Developer Experience

• Enhanced CLI with examples, defaults, and environment variables
• Pre-commit hooks with Ruff, mypy, bandit, and more
• Ruff configuration with 25+ enabled rule sets
• PEP 561 py.typed marker for type checking
• JSON schemas for scenarios, results, metrics, and datasets

Community & Project Management

• GitHub issue templates for bugs, features, docs, and questions
• Pull request template with comprehensive checklist
• 50+ organized labels for project management
• Hacktoberfest-ready tags and labels
• Example CI workflow showing AgentUnit integration

Quality Improvements

• Fixed Python version mismatch (now consistently 3.10+)
• Added CI badges to README
• Copy-paste quickstart examples
• Optional extras documentation
• 144 passing tests

Installation

Basic: pip install agentunit
All extras: pip install agentunit[all]
Specific: pip install agentunit[langraph,crewai,openai]

Breaking Changes

None - fully backward compatible with v0.5.0

Full Changelog

See CHANGELOG.md for complete details

v0.5.0

Release v0.5.0

Changelog:

• Bump package version to 0.5.0
• Documentation rewrite and improvements
• RegressionDetector boolean fix; no-code builder adjustments

All tests passing locally: 144 passed, 10 skipped.

See the commit for details: b3d794a

v0.4.0

🎉 Release v0.4.0

✨ Added

• PEP 562 lazy loading implementation for optional framework dependencies
• Comprehensive TYPE_CHECKING imports for static analysis and IDE support
• Graceful degradation for telemetry when OpenTelemetry is not available
• Enhanced dataset registry with optional HuggingFace Hub support
• .codacy.yml configuration for linter-friendly lazy loading pattern
• OPTIONAL_DEPENDENCIES.md comprehensive documentation guide
• New adapters: AgentOps, LangSmith, Wandb, Swarm, AG2
• CLI module with command-line interface
• Multi-agent coordination framework
• Production integration utilities

🔄 Changed

• Refactored src/agentunit/adapters/init.py to use getattr for lazy imports
• Updated src/agentunit/telemetry/tracing.py with no-op fallbacks
• Enhanced module docstrings documenting lazy loading pattern
• Bumped project version to 0.4.0

🐛 Fixed

• ModuleNotFoundError cascades from optional dependencies
• Runtime import errors when optional packages not installed
• Adapter availability checks now work correctly without all dependencies

📚 Documentation

• Added comprehensive guide for optional dependencies
• Enhanced API reference documentation
• Improved architecture documentation
• Added performance testing guides

🔗 Links

• Full Changelog
• Documentation
• Optional Dependencies Guide

🙏 Contributors

Thank you to all contributors who made this release possible!

v0.3.0

v0.3.0 (2025-09-30) - Documentation & Framework Expansion

🎯 Highlights

• Major Documentation Overhaul: Expanded scenario helper documentation covering 9 new frameworks
• Framework Integration Catalog: Comprehensive guide with prerequisites, helper signatures, and customization strategies
• Developer Templates: New framework scenario template (framework_scenarios.py) with 9 helper examples

📚 Added Frameworks & Adapters

• Phidata - Data-centric agent helpers
• PromptFlow - Microsoft PromptFlow orchestration scenarios
• OpenAI Swarm - OpenAI Swarm multi-agent scenarios
• Anthropic on Bedrock - Claude integration helpers
• Mistral Server - Open-source Mistral deployment scenarios
• Rasa - Conversational AI agent helpers
• AutoGen - Conversational orchestration adapters
• Haystack - Pipeline-based AI scenarios
• LlamaIndex - Query engine integration helpers

🔧 Infrastructure

• Enhanced adapter registry with registration and resolution utilities
• Comprehensive test coverage for all new adapters
• Updated documentation architecture with integration guides

📦 Maintenance

• Version bump to 0.3.0 across all package metadata
• Added structured CHANGELOG.md for release tracking

---

🚀 Initial Production Release

Complete package setup ready for PyPI publication with enterprise-grade features and documentation.

✨ Core Features

• Comprehensive Testing Framework: Full scenario-based testing for AI agents
• Multi-Framework Support: Built-in adapters for CrewAI, LangGraph, and OpenAI Agents
• Built-in Datasets: Curated test datasets for common AI scenarios
• Metrics & Reporting: Extensible metrics system with detailed result reporting
• CLI Interface: Command-line tools for running test suites and managing scenarios

🏗️ Core Architecture

• Scenario Engine: Flexible scenario definition and execution system
• Adapter Pattern: Pluggable adapters for different AI frameworks
• Trace Management: Comprehensive execution tracing and replay capabilities
• Dataset Registry: Extensible dataset management system
• Metrics Framework: Built-in and custom metric evaluation system

📚 Documentation Suite

• README: Professional package overview with installation and usage
• Quickstart Guide: Step-by-step getting started tutorial
• CLI Reference: Complete command-line interface documentation
• Scenario Writing Guide: Comprehensive guide for creating test scenarios
• Templates: Ready-to-use suite templates with examples

🔧 Developer Infrastructure

• GitHub Actions: Automated PyPI publishing workflow
• Security Setup: Comprehensive credential management documentation
• Build System: Poetry-based dependency management
• Testing: pytest-based test suite with fixtures
• Publishing: Automated release workflow with security best practices

📦 Package Metadata

• License: MIT License for open-source adoption
• Python Support: Python 3.8+ compatibility
• Dependencies: Carefully curated dependency list for stability
• Entry Points: CLI commands via agentunit command

🛠️ Built-in Components

• Adapters: CrewAI, LangGraph, OpenAI Agents
• Datasets: Customer support, code review, research scenarios
• Metrics: Response quality, accuracy, performance metrics
• Reporting: JSON, console, and structured result formats
• Examples: Template project with working examples

🔐 Security & Best Practices

• Environment variable configuration for sensitive data
• Secure credential management documentation
• Protected publishing workflow with environment gates
• Comprehensive .gitignore for secure development

---

Version Summary

Version	Release Date	Type	Key Focus
v0.3.0	2025-09-30	Feature	9 new framework adapters, documentation expansion