Use the Lens built-in browser to inspect a running web application. Trigger when the request involves previewing a page, taking screenshots, reading DOM, checking console errors, picking elements to locate source code, or comparing rendered output to a design spec.
86
82%
Does it follow best practices?
Impact
90%
1.57xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Inspect and interact with a live web application through Stave's built-in browser panel.
Lens is a per-workspace embedded browser in the right rail panel. It renders pages via Electron's WebContentsView and exposes 10 MCP tools so AI agents can navigate, screenshot, read DOM, evaluate JavaScript, and interact with page elements programmatically.
All tools require a workspaceId parameter.
| Tool | Purpose |
|---|---|
stave_lens_navigate | Load a URL in the browser |
stave_lens_screenshot | Capture viewport or full-page PNG |
stave_lens_get_html | Get outerHTML of page or a selector |
stave_lens_get_text | Get textContent of a selector |
stave_lens_evaluate | Run JavaScript in page context |
stave_lens_get_console | Read buffered console messages (up to 200) |
stave_lens_get_network | Read buffered network requests (up to 200) |
stave_lens_click | Click an element by CSS selector |
stave_lens_type | Type text into a focused or specified element |
stave_lens_snapshot | Get accessibility tree snapshot |
The crosshair button in the address bar activates the element picker.
Captured data is inserted into the active task's prompt draft:
_debugSource file:line (when enabled in Settings)Configure in Settings > Lens > Source Code Mapping.
The AI uses distinctive class names, text content, IDs, and component-style class patterns from the picked element to search your codebase with grep or file-search tools. Works with any framework.
When your application runs in React dev mode with @babel/plugin-transform-react-jsx-source (enabled by default in Vite React plugin, CRA, and Next.js dev), the picker extracts the exact fileName:lineNumber from React fiber internals. This gives 100% accurate source location when available, with automatic fallback to heuristic search.
Combine Lens with a Figma MCP server for design-to-code verification:
1. figma-mcp: read design spec (colors, spacing, layout)
2. stave_lens_navigate("http://localhost:3000/page")
3. stave_lens_screenshot()
4. Compare rendered output to Figma spec
5. Modify code to fix discrepancies
6. Repeat from step 2Page.getLayoutMetrics to capture beyond the viewport.stave_lens_snapshot) give a compact page structure summary — useful when full HTML is too verbose.file:// or chrome:// protocols — they are blocked for security._debugSource in production builds — it's only available in dev mode.about:blank yields an empty image.e275ad5
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.