Evaluation Simulations (Single Run)

Execute a single run with full per-step logging for debugging

Evaluation Simulations (Single Run)

An Evaluation run executes your simulation once with full per-step logging. Use it while you're building and debugging — you can step through every decision, watch datapoints change, and see exactly what each Behavior node picked.

Setting up the run

Open the Evaluation tab on the simulation page. The Runner shows:

  • Game Version — the version the simulation runs against. Newly created or unpublished versions work too.
  • Steps Limit — a hard cap on the number of steps before the simulation auto-stops (clamped between 10 and 2000). Raise this for long-running loops.
  • Seed — optional. If blank, a seed is generated for you and shown in the results. Reuse a seed to reproduce the same run byte-for-byte.
  • Profiles — pick a single profile to use its variable values (or leave none selected to use the current slider values). See Profiles.

Running

Click Run Simulation. The backend executes your node graph once, recording every step:

  • Which node was executing
  • Which outcome a Behavior node picked
  • Which datapoints changed and by how much
  • The derived seed used for each randomized decision

What you get back

  • Execution log — the full list of executed steps, in order.
  • Datapoint series — the value of each datapoint at every step, ready to chart.
  • Seed — the actual seed used (whether you provided one or not).

The simulation automatically opens the Step Insights panel so you can start scrubbing right away — see Step Insights.

When to use Evaluation

  • You're iterating on the graph and want to see exact decisions
  • Something is misbehaving and you need to pinpoint where
  • You want a reproducible run (pin the seed and share)

For questions about variance ("what's the average gold at step 100?"), you want Statistical Simulations instead.

Next

Continue to Step Insights to learn the step-by-step debugger.