git clone https://github.com/daniellopez882/agentbench.git
cd agentbench
pip install -e ".[dev,llm]"
pytest tests/ -v- Built-in scenario packs — Add YAML scenarios for common agent types
- New evaluators — trajectory efficiency, plan quality, tool accuracy
- Framework adapters — pre-built adapters for LangGraph, CrewAI, AutoGen
- HTML report generator — standalone report with charts
- Trace capture — record tool calls and intermediate reasoning steps
- Python 3.10+ with type hints
- Lint with
ruff check - Tests required for new features
- Every evaluator needs at least pass + fail test cases