Agent System Simulator v1.0
A runnable multi-agent workflow simulator demonstrating governance, orchestration, and evaluation through working code.
What's included
- Four agent roles: Planner, Executor, Evaluator, Supervisor
- Three demonstration scenarios: normal_success, retry_then_success, fallback_after_failure
- Decision logging, retry/escalation tracking, latency and cost estimation
- Evaluation summary on each run
Quick start
pip install -r requirements.txt
python run_demo.py --scenario normal_successMIT License.