Benchmark harness on capability questions
hle-15-ablationworkflowsetup L3★4
ejentum/benchmarks ↗What it does
Orchestrate multi-stage agent work
Best for
Executing complex multi-phase workflows with typed outputs.
Outputs
- · Typed structured output
Requires
- · agent() function
- · phase() phase-tracking
Preconditions
- · Workflow runtime initialized
- · Input arguments validated
Failure modes
- · Gate validation fails
Trust signals
- · Binary/enum verdict logic in code