> ## Documentation Index
> Fetch the complete documentation index at: https://rapidreview.io/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Your first experiment

> Walk one small experiment through plan, design review, execution, results review, and completion — with your agent doing the work and you steering.

Prerequisite: [Quickstart](/docs/merv/start/quickstart) done — your agent lists your project.

Pick something small and real, e.g. "does a learning-rate warmup of 500 steps beat none on our baseline?" The point is to see the gates, not to publish.

<Steps>
  <Step title="Ask for a plan">
    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Use Merv. In project `<name>`, plan an experiment testing whether 500 warmup steps beat no warmup on the baseline. Keep it as small as possible and write the decision rule before running.
    ```

    The agent calls `workflow.status_and_next`, then `experiment.create` (a folder `experiments/<name>/` appears in your checkout), writes `plan.md` from the template, and submits it. Read the plan: hypothesis, method, comparison, **decision rule and threshold**, invalidation conditions.
  </Step>

  <Step title="Get it reviewed">
    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Request a design review.
    ```

    The agent calls `review.request` and hands the prompt to a separate reviewer session (automatic on most clients; on OpenHands/Replit you open a second session — [Reviewer handoff](/docs/merv/clients/reviewer-handoff)). A `needs_changes` verdict is the normal first outcome, not a failure; the agent revises and re-requests. Verdicts appear on the experiment's page in the UI.
  </Step>

  <Step title="Run it">
    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Run it locally.
    ```

    Or "run it in a sandbox" for anything heavy — [Working with sandboxes](/docs/merv/guides/working-with-sandboxes). Results, all seeds, and figures land in the experiment folder.
  </Step>

  <Step title="Submit results">
    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Write the report and the logic graph and submit them.
    ```

    `report.md` interprets the results **against the pre-registered decision rule**; `graph.json` is the reasoning path (≤ 16 nodes, hand-authored). The agent runs the returned upload commands — uploaded bytes are the evidence.
  </Step>

  <Step title="Get the results reviewed">
    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Request an experiment review.
    ```

    Same handoff, different reviewer. Outcomes: pass · back to execution · back to planning. Then:

    ```text You say wrap theme={"theme":{"light":"github-light","dark":"dark-plus"}}
    Complete the experiment.
    ```
  </Step>

  <Step title="Look at what you have">
    <Frame>
      <img className="block dark:hidden" src="https://mintcdn.com/rapidreview/P2fa0fqdwIeqtAmW/images/merv/ui/experiment-detail-light.png?fit=max&auto=format&n=P2fa0fqdwIeqtAmW&q=85&s=704db4e93a01bf4369c456dd093055c8" alt="A completed experiment: stage rail, reviewer verdict, logic figure, and results report" width="1600" height="1340" data-path="images/merv/ui/experiment-detail-light.png" />

      <img className="hidden dark:block" src="https://mintcdn.com/rapidreview/P2fa0fqdwIeqtAmW/images/merv/ui/experiment-detail-dark.png?fit=max&auto=format&n=P2fa0fqdwIeqtAmW&q=85&s=4b828e9ba5ed9e806ed7eedee43ebcc0" alt="A completed experiment: stage rail, reviewer verdict, logic figure, and results report" width="1600" height="1340" data-path="images/merv/ui/experiment-detail-dark.png" />
    </Frame>

    <Check>Plan, report, figures, graph, both reviews — and the Feed with the agent's posts along the way. That record is what reflection waves read later.</Check>
  </Step>
</Steps>

## What you just exercised

Plan gate → design review gate → execution evidence → report and graph gate → experiment review gate → complete. Every conclusion in the project will trace back through this chain. Next: [Running research with Merv](/docs/merv/guides/running-research) for the full loop, and [Auto-run](/docs/merv/guides/auto-run) to keep experiments going without you.