# Web Test: An Agentic Web Automation Testing Solution

Web Test in Agentic Mode lets you automate web journeys using plain English. You describe what you want to validate, and the agent opens the site, explores it visually, converts your intent into a step-by-step plan (actions + assertions), and then executes the flow while streaming live screenshots. It finishes with a detailed report containing outcomes, evidence, and logs—so the result is both clear and traceable.

<figure><img src="/files/5aHiqslyjvF52mAKElEn" alt=""><figcaption></figcaption></figure>

### **Key Benefits**

* **Review and edit before execution**\
  The agent proposes steps and assertions first. You can modify them—add, remove, or refine—before running the test.
* **Live visual trace**\
  Watch the execution as it happens with step-by-step screenshots and a timeline that supports click-to-jump navigation.
* **Evidence-backed assertions**\
  Each validation is tied to screenshots and logs, so pass/fail outcomes are explicit and explainable.
* **Rerun with changes**\
  Iterate quickly by adjusting steps or assertions and rerunning—no need to rebuild the entire test.
* **DeepProbe insights (optional)**\
  Include accessibility checks and web performance metrics directly in the report.
* **Export and share**\
  Download a PDF report and share results easily.
* **CI/CD integration**\
  Trigger web tests from your pipeline, poll for completion, and attach reports as build artifacts to gate releases.

### **Preconditions**

Before running a **Web Test,** make sure:

* The target website is reachable and behaving normally.
* Your scenario is clear and unambiguous.
* Any required test data is ready (CSV, if you’re running bulk tests).

### How to Run a Web Tes&#x74;**?**

#### **Step 1:** Start a new test with a scenario

<figure><img src="/files/4mN1VWVzD1bF6OTQe5RP" alt=""><figcaption></figcaption></figure>

Go to the **Web Test** section and provide:

* A clear natural-language prompt (what to do + what to validate)
* The target URL
* Optional test data (CSV upload for bulk runs)

You can also pick from suggested templates, such as:

* Add-to-cart flow on Swadesh
* Video playback test on MX Player
* Flight discovery on Cleartrip

#### **Step 2:** Review the plan (optional but recommended)

<figure><img src="/files/arSf1xwxNe70XGNNkPbS" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/BREeLR9tSGAtdwX7gR1S" alt=""><figcaption></figcaption></figure>

* If you choose the detailed planning option, the agent will generate a proposed plan with steps and assertions.\
  \
  You can then:
  * **Accept :** proceed with the proposed plan

<div align="center"><figure><img src="/files/Ol4Alwhinhkm4gVfhptF" alt="" width="375"><figcaption></figcaption></figure> <figure><img src="/files/IZZBDXaENIfsXQWo2F3n" alt="" width="375"><figcaption></figcaption></figure></div>

* **Modify :** adjust steps or assertions as needed, then submit

<div align="left"><figure><img src="/files/VVRpocxumACgYUEFybmB" alt="" width="375"><figcaption></figcaption></figure> <figure><img src="/files/C55NXgWdMrppr9lbVlOY" alt="" width="375"><figcaption></figcaption></figure></div>

Once the plan is submitted—or if you skip planning—you’ll land on the execution screen.

<figure><img src="/files/cD8X3u8lSxNUqcjblow7" alt=""><figcaption></figcaption></figure>

#### **Step 3: Visualise the execution**

During execution, the agent:

* Launches the URL in a real browser
* Follows the approved step plan in order
* Runs assertions after each step and marks pass/fail
* Captures step-level screenshots as evidence
* Streams updates to the live timeline until completion

<div align="left"><figure><img src="/files/58StnpYH9998q9VVhBBl" alt="" width="375"><figcaption></figcaption></figure> <figure><img src="/files/ZOcOoTmfKzwy6GuaS6Bq" alt="" width="375"><figcaption></figcaption></figure></div>

#### **Step 4 : Analyse the Report**

What the report includes

* **Stats**: total assertions, passed, failed.
* **Summary**: AI‑generated overview of the run outcome and intent.
* **Overview**: URL, triggered by, timestamps, duration, browser, viewport.
* **Step Sequence**: The exact step plan (actions + assertions).
* **Assertions table**: expected vs actual, pass/fail, linked screenshots.
* **Screenshots gallery**: full visual trace of the run.
* **Test logs:** step‑level logs and outcomes.
* **Export assets**: Video Recording + PDF export.
* **DeepProbe (if enabled)**: accessibility + web performance results.
* **Network + console logs (if DeepProbe enabled)**: captured during execution.

<figure><img src="/files/YVJjiCTBNAToRCPYtvMM" alt=""><figcaption></figcaption></figure>

### **Advanced Features**

#### **Bulk Test Execution**

* Upload a CSV with multiple data rows along with your prompt. The system generates appropriate test cases, runs them in parallel, and provides individual reports for each run.

<div><figure><img src="/files/yjxa5OAQtdriY9ydkIHI" alt=""><figcaption></figcaption></figure> <figure><img src="/files/O91pxFoceBUNQXcl42vI" alt=""><figcaption></figcaption></figure></div>

<figure><img src="/files/M1dO6B7jw0I2e42zqQ6c" alt=""><figcaption></figcaption></figure>

#### **Test Management**

* View all test executions in a centralised dashboard
* Filter and search tests by name, status, or date
* Re-run previous tests with a single click
* Access detailed execution history

<figure><img src="/files/o8BH9SBSdtFMI1qbDbXM" alt=""><figcaption></figcaption></figure>

#### **Recording and Playback**

* Watch video recordings of test executions
* Review step-by-step screenshots
* Analyze execution flow for debugging

<figure><img src="/files/ap5iWWFJidqLmGtQNNdG" alt=""><figcaption></figcaption></figure>

### **Postconditions**

After test execution:

* A full report is generated
* A video recording is available
* Step-by-step screenshots are captured as evidence
* Final status (success/failure) is clearly shown
* Key execution metrics (assertion counts, pass/fail, duration) are summarized

### **Best Practices**

1. **Writing strong instructions**

* Be specific about actions (click, type, select, navigate)
* Include expected outcomes in the prompt
* Break complex flows into clear, logical steps<br>

2. **Managing test data**

* Use CSV for bulk execution
* Keep formats consistent
* Include varied data to improve coverage<br>

3. **Monitoring and improving runs**

* Review reports after each execution
* Use failed steps to refine prompts and assertions
* Track success rates over time to measure stability

### **Conclusion**

Web Test in ratl.ai turns natural-language intent into reliable, vision-driven web automation. It removes scripting by generating executable steps and assertions, provides live visual evidence during runs, and closes with a comprehensive report—making web testing faster to create, easier to debug, and simple to share.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.ratl.ai/guide/features/web-automation.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.