> ## 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.

# Critiqor FAQ: Installation, Privacy, and Usage Questions

> Answers to the most common questions about Critiqor — from what it does and which frameworks it supports to data privacy and contributing.

Got questions about Critiqor? Here are answers to the things developers ask most often — covering what Critiqor does, how it handles your data, and how to get the most out of it.

<AccordionGroup>
  <Accordion title="What is Critiqor?">
    Critiqor is an **AI Agent Runtime Intelligence Platform**. It observes your AI agents during execution, captures runtime evidence, and delivers reliability diagnostics through a local dashboard.

    It helps you see what your agent *actually did* — not just what it reported. Instead of relying on logs or agent self-reporting, Critiqor gives you an independent record of what happened during a real run.
  </Accordion>

  <Accordion title="Which frameworks are supported?">
    Critiqor currently supports **OpenClaw**. You can monitor any agent running through OpenClaw using `critiqor monitor openclaw`.

    Support for additional frameworks and runtimes is on the roadmap. If you'd like to request a specific integration, open a GitHub issue at [github.com/web3curtis/Critiqor](https://github.com/web3curtis/Critiqor).
  </Accordion>

  <Accordion title="Is my data private?">
    Yes. Critiqor stores all run artifacts **locally on your machine** by default. Nothing leaves your machine unless you have explicitly configured a backend connection via `CRITIQOR_BACKEND_URL`.

    You control when observation starts, when it ends, and which results you review or share. Critiqor does not send telemetry or usage data without your explicit configuration.
  </Accordion>

  <Accordion title="Does Critiqor modify my agent?">
    No. Critiqor is a **passive observer**. It does not intercept, modify, or change your agent's responses or behavior in any way. It runs alongside your agent and records what happens — your agent behaves exactly as it would without Critiqor present.
  </Accordion>

  <Accordion title="How do I update Critiqor?">
    Run:

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

    Check your current version at any time with:

    ```bash theme={null}
    pip show critiqor
    ```
  </Accordion>

  <Accordion title="Can I evaluate multiple runs?">
    Yes. Every `critiqor finalize` creates a new run with its own reliability report. Your runs accumulate over time, so you can track how your agent performs across different sessions or code changes.

    List all your runs:

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

    Open a specific run's dashboard:

    ```bash theme={null}
    critiqor dashboard run_001
    ```
  </Accordion>

  <Accordion title="Can I use Critiqor in CI?">
    Yes. Use `critiqor check` to run policy checks against your evaluation history. It exits **non-zero** if your defined thresholds are not met, making it compatible with CI/CD pipelines like GitHub Actions, GitLab CI, and CircleCI.

    ```bash theme={null}
    critiqor check   # exits 0 if all policies pass, non-zero otherwise
    ```

    See the [Policy Checks guide](/guides/policy-checks) for details on configuring thresholds and failure conditions.
  </Accordion>

  <Accordion title="Does Critiqor replace tests?">
    No. Critiqor **complements** your existing test suite — it doesn't replace it.

    Unit tests, integration tests, evals, and human review each catch different classes of problems. Critiqor adds a layer that shows you what actually happened during a real agent run, which is something static tests alone can't provide. Use all of these together for the most reliable agent development workflow.
  </Accordion>

  <Accordion title="Where can I report bugs?">
    Open a GitHub issue at [github.com/web3curtis/Critiqor](https://github.com/web3curtis/Critiqor).

    To help us resolve it quickly, include:

    * Your Critiqor version (`pip show critiqor`)
    * Your Python version (`python --version`)
    * The exact command you ran
    * What you expected to happen
    * What actually happened (include any error output)
  </Accordion>

  <Accordion title="How do I contribute?">
    Useful contributions include:

    * **Bug reports** — clear, reproducible issues with version and environment details
    * **Documentation improvements** — corrections, clarifications, or missing examples
    * **OpenClaw workflow feedback** — real-world usage patterns that could improve the experience
    * **Dashboard usability feedback** — report anything confusing or missing in the local dashboard
    * **Framework integration requests** — let us know which runtimes or agent frameworks you'd like supported

    If you're proposing a larger change — such as a new feature or a significant refactor — open an issue first to discuss it before writing code. This ensures your effort aligns with the project's direction.
  </Accordion>

  <Accordion title="What Python versions are supported?">
    Critiqor supports **Python 3.10, 3.11, 3.12, and 3.13**.

    To check which version you're running:

    ```bash theme={null}
    python --version
    ```

    If you're on Python 3.9 or earlier, upgrade to a supported version before installing Critiqor.
  </Accordion>
</AccordionGroup>