#Workflow Benchmark Corpus

    This directory is a deterministic scenario benchmark for moon-fsm. It is designed to make the project scope and behavior reviewable without depending on wall-clock timing, CPU model, or a network service.

    The checked-in CSV contains 44 representative workflow cases across four common engineering domains:

    • approval and review workflows
    • order fulfillment and cancellation
    • device boot, fault, and recovery flows
    • support ticket assignment, escalation, reopening, and closure

    The extended support cases also cover customer waiting, cancellation, terminal-event storms, and long reopen loops. The runnable examples extend those scenarios with examples/support_workflow for customer operations and examples/incident_workflow for incident commander assignment, mitigation, rollback, journal ingestion, SLA checks, runbook handover, and closure review.

    Each case records the initial state, initial context, event sequence, expected final state, successful transition count, and rejected-event count. The executable runner in main.mbt uses the same cases as typed data and verifies the expected history length and audit-entry count after every scenario.

    Run the benchmark from the module root:

    moon run benchmarks

    This is a behavioral scenario benchmark, not a claim about throughput across hardware. It complements the unit tests by exercising longer, realistic paths and explicit failure/recovery branches.

    Scenario

    type Scenario