> ## Documentation Index
> Fetch the complete documentation index at: https://critiqor-71f5274a.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart: Run Your First Critiqor Agent Evaluation

> Observe your first OpenClaw agent run with Critiqor, finalize the session, open the dashboard, and review your reliability report in five steps.

This page walks you through your first complete Critiqor workflow — from installing the CLI to reviewing your first reliability report. By the end, you'll have captured a live agent session and have actionable insights about how your agent behaved in the real world.

<Steps>
  <Step title="Install Critiqor">
    Install the Critiqor CLI using pip:

    ```bash theme={null}
    pip install critiqor
    ```

    Verify the installation by running:

    ```bash theme={null}
    critiqor help
    ```

    You should see the list of available commands. If the command isn't found, ensure your Python environment's `bin` directory is on your `PATH`.
  </Step>

  <Step title="Start an observation session">
    Launch an observation session for your OpenClaw agent:

    ```bash theme={null}
    critiqor monitor openclaw
    ```

    Critiqor starts capturing runtime evidence and then launches OpenClaw in the same terminal. You'll see the OpenClaw interface exactly as you normally would — no changes to your workflow are required.
  </Step>

  <Step title="Use OpenClaw normally">
    Work with your agent as you usually would. Ask it questions, assign it tasks, and let it run. Critiqor operates silently in the background, recording what your agent actually does — including tool calls, retries, timing, and any errors that occur.

    There's nothing special you need to do during this step. The more representative your usage, the more useful your reliability report will be.
  </Step>

  <Step title="Finalize the session">
    When you're done, finalize the observation session:

    ```bash theme={null}
    critiqor finalize
    ```

    Critiqor completes the observation, processes all captured runtime evidence, generates your reliability report, and automatically opens the dashboard in your default browser.
  </Step>

  <Step title="Review the dashboard">
    The dashboard displays your full reliability report. Here's what each section tells you:

    * **Executive Summary** — A plain-language overview of how your agent performed during the session.
    * **Trust Score** — A high-level reliability rating that reflects the overall quality of your agent's behavior.
    * **Primary Diagnosis** — The most significant reliability finding from the session, with context on why it matters.
    * **Runtime Evidence** — The raw behavioral data captured during observation, including tool calls, retries, timing, and errors.
    * **Recommendations** — Concrete, prioritized steps you can take to improve your agent's reliability.

    For a full walkthrough of every dashboard section, see the [Dashboard Guide](/dashboard/overview).
  </Step>
</Steps>

<Note>
  All session runs are stored locally on your machine. You can reopen any previous report at any time by running `critiqor dashboard` to view the most recent session, or `critiqor runs` to browse and select from all past sessions.
</Note>

## Next steps

Now that you've run your first evaluation, explore the rest of the Critiqor docs to get more out of every session.

<CardGroup cols={2}>
  <Card title="Dashboard Guide" icon="chart-bar" href="/dashboard/overview">
    Explore every section of the reliability dashboard in depth.
  </Card>

  <Card title="CLI Reference" icon="terminal" href="/cli/overview">
    View all available Critiqor CLI commands and their options.
  </Card>

  <Card title="Observation Workflow" icon="eye" href="/guides/observation-workflow">
    Learn how to get the most out of your observation sessions.
  </Card>

  <Card title="Best Practices" icon="star" href="/guides/best-practices">
    Follow recommended patterns for reliable, repeatable agent evaluation.
  </Card>
</CardGroup>