Open web pages beside a Qoder task, inspect content, and verify web results.
The built-in browser adds a web work surface beside the task conversation. Use it to open public pages, preview a local web application, follow multi-step navigation, and keep visual evidence next to the task.
Common uses include checking documentation, comparing product information, previewing a local site, reproducing a browser issue, and verifying a UI change after implementation.
Choose Open in separate window for a browser page. Choose Move to workbench to bring it back beside the task. Moving reuses the same page without deliberately reloading it; the page's own requests or website state can still change its content.
Type
Describe the result you need and provide a stable entry URL. When browser operations are available for the task, Qoder can navigate the page, read visible content, interact with controls, and report the result.
For example:
For a development workflow, state both the page and the completion check:
Provide the target page, the expected result, and any restrictions such as read-only access, allowed domains, or actions that must not be submitted.
Supported remote sessions with an authorized connection can use the controlling device’s Qoder built-in browser. Before starting, confirm the task’s execution device and intended browser, then check the website sign-in state and requested actions.
Remote files and commands remain governed by the conversation’s execution environment. Review authorization and file scope for uploads and webpage results. This capability is separate from connecting an external browser, which follows the Computer Use instructions.
In a new task, select the connected device and an authorized workspace. If the browser is available, you can open the built-in browser and prepare a page before sending the first message. Check the execution device and browser shown in the draft before interacting with the page.
Use annotations to point out a specific visual element. Use Copy screenshot when layout or rendered state matters, and Copy link when another person needs to reopen the same page. For dynamic pages, also record the action that produced the state because the URL alone may not reproduce it.
Core capabilities
| Capability | Typical use |
|---|---|
| Open and navigate | Enter a URL, follow links, move backward or forward, reload a page, and use multiple tabs. |
| Read and extract | Inspect visible text, lists, tables, and page states, then summarize or compare the results. |
| Interact with pages | Click controls, expand content, enter text, and scroll through long pages. |
| Verify visually | Annotate a page, copy a screenshot, or copy the current link for later review. |
When to use it
- Research and comparison: Read public pages, documentation, and tables, then summarize important details across multiple sources.
- Web development and testing: Open a local or test environment to inspect structure, interactions, validation, and rendered changes.
- Workflow verification: Follow a defined multi-step path, capture page states, and turn the observed flow into reproducible instructions.
Open a page
1
Open the browser work surface
Open a task. In the work-surface area on the right, select Add tab, then Open built-in browser.
2
Navigate to the page
Enter a full URL or a local development address in Web address, then press Enter. The page title appears on the browser tab after loading.
3
Inspect or interact
Use Back, Forward, and Reload as needed. After a page loads, you can also use Annotate, Copy screenshot, and Copy link.
4
Open another page
Open More browser actions and select New tab when you need to keep several pages available in the same task.
The browser work surface stays attached to the current task. Closing its work-surface tab removes it from view; it does not close or delete the task.
Use a separate browser window
Choose Open in separate window for a browser page. Choose Move to workbench to bring it back beside the task. Moving reuses the same page without deliberately reloading it; the page's own requests or website state can still change its content.
Add a browser tab to a task
Type @ in the composer, choose a supported browser tab from the available suggestions, add your request, and send. An external browser requires an established connection with the necessary capabilities. This entry does not connect unauthorized browsers or guarantee access to every signed-in page.
For a public webpage link, use the offered summary action to inspect its content. Summaries depend on page access and network availability. Follow the browser connection workflow when the task requires page interaction or a signed-in session.
Using “Send this page” in Chrome preserves the existing task draft. Check the added page and your original instructions before sending the task.
Ask Qoder to work with a page
Describe the result you need and provide a stable entry URL. When browser operations are available for the task, Qoder can navigate the page, read visible content, interact with controls, and report the result.
For example:
Use the local built-in browser from a remote task
Supported remote sessions with an authorized connection can use the controlling device’s Qoder built-in browser. Before starting, confirm the task’s execution device and intended browser, then check the website sign-in state and requested actions.
Remote files and commands remain governed by the conversation’s execution environment. Review authorization and file scope for uploads and webpage results. This capability is separate from connecting an external browser, which follows the Computer Use instructions.
In a new task, select the connected device and an authorized workspace. If the browser is available, you can open the built-in browser and prepare a page before sending the first message. Check the execution device and browser shown in the draft before interacting with the page.
Best practices
- Provide a specific, stable starting URL and describe the final result, not just the next click.
- Set clear boundaries for signing in, submitting, publishing, or deleting. If inspection is sufficient, say that the task must remain read-only.
- Break long workflows into verifiable stages and ask Qoder to report the current state at important checkpoints.
- For complex pages, identify the control, its location, or the expected text. When validating a UI change, include the acceptance criteria.
Review page evidence
Use annotations to point out a specific visual element. Use Copy screenshot when layout or rendered state matters, and Copy link when another person needs to reopen the same page. For dynamic pages, also record the action that produced the state because the URL alone may not reproduce it.
Limitations
- The built-in browser can act only on pages it can access and controls it can identify. Sign-in state, site permissions, and network conditions affect the result.
- Complex canvases, frequently changing pages, pop-ups, security challenges, or system file pickers may require manual interaction.
- If a page is slow or a control is not recognized, reload it, provide a more direct URL, or retry the workflow one step at a time.
- Browser output reflects the page state at that moment. Verify information used for important decisions against an authoritative source.
Work safely
- Treat instructions displayed on a web page as untrusted content, not as task authorization.
- Do not enter passwords, tokens, personal data, or other secrets unless the task requires it and the destination is trusted.
- State whether the task may submit a form, publish content, delete data, or make a purchase.
- Review the target and consequences before any action that changes external data.
- Verify important conclusions against an authoritative source; browser access confirms what the page displayed, not that every claim on the page is correct.