SiteLens · Browser Intelligence
Run a real browser locally for free, or capture and extract data from public websites using the SiteLens Cloud API.
SiteLens Local and SiteLens Desktop run on your machine. SiteLens Cloud API runs on our infrastructure and reads public webpages. Cloud capture does not open localhost, sign in, or click through a flow.
Free MCP tools and CLI. No account or API key. Automate a real browser on your machine, including localhost and private apps, with saved authenticated sessions.
A visual interface for the same local engine, on macOS Apple Silicon. Included on Free and Pro.
Hosted capture of public pages: HTTP, HTML, or a rendered browser with a screenshot, DOM, and rendered text. Author locators visually or by description, then call them back with POST /capture.
Find the data you need. Visually or by description. Preview a public page, click a price, a product detail, or an availability line — or describe it, for example “Professional plan monthly price” — then confirm the match and copy a reusable POST /capture request. Matching is deterministic. SiteLens returns the current content on each request; it does not monitor changes, store extraction configurations, or send notifications. Nothing is installed on the page.
Click the element on a rendered preview, the way you would in a browser inspector. Review the value, then keep the locator.
Ask for a plan price, a product name, or whether an item is in stock. SiteLens ranks candidates from the page. You confirm the match.
Name up to 20 targets — starter price, product details, availability — and read them from a single POST /capture call. Your application keeps the request.
Send a public URL. SiteLens retrieves the page from our infrastructure and returns the evidence in the response. You do not install a script, change DNS records, or grant access to the site’s servers. Private, localhost, and signed-in apps stay on your machine with SiteLens Local or Desktop.
Response headers and body. No browser. Useful when the document you want is already in the HTTP response.
Fetched source, plus text, metadata, and links parsed from that source. Still no browser, unless you ask for a browser fallback.
The default. A real browser run returns a full-page screenshot, rendered text, and the DOM after JavaScript. This is the mode for prices and stock lines that are not in the original HTML.
POST https://api.sitelensapi.com/capture
Authorization: Bearer sitelens_sk_…
{ "url": "https://example.com", "mode": "http" }
{ "url": "https://example.com", "mode": "html" }
{ "url": "https://example.com", "mode": "browser" }HTTP returns headers and the body. HTML adds text, metadata, and links from that source. Browser, the default, returns a full-page PNG, rendered text, and the DOM after JavaScript. None of these modes logs in or clicks.
{
"url": "https://example.com/pricing",
"mode": "browser",
"locators": {
"price": { "css": ".plan-pro .price" }
},
"targets": [
{ "name": "price", "locator": "price", "extract": "text" }
]
}Click a preview or describe a field such as a price, product detail, or availability line, then confirm the match. Later requests send the locator, not the description. SiteLens does not store the recipe or watch the page.
No. For a public webpage, SiteLens calls the URL from our infrastructure. There is no script to add, no DNS record to change, and no server access to grant. You need a SiteLens API key on your side, not on the site you are reading. The URL has to be a public HTTP or HTTPS page. Localhost, private networks, and signed-in sessions are outside this path — use SiteLens Local or Desktop on a machine that can already reach them.
HTTP returns headers and the response body. HTML adds text, metadata, and links parsed from that source. Browser capture, the default, loads the page and returns a full-page PNG, the rendered text, and the DOM. Screenshots and JavaScript-rendered content require browser mode. HTTP and HTML can fall back to a browser when you set fallback to browser.
Both are ways to author a capture in the dashboard, not a second product. Visual selection is a click on the preview. Description-based extraction is a phrase such as “Professional plan monthly price,” “product details,” or “in stock.” SiteLens ranks likely elements with deterministic text and structure matching, and you confirm one. The copied POST /capture request uses ordinary locators — CSS, test id, role and name, or text — not the description. A later call returns the current values. SiteLens does not store the configuration or watch the page for changes.
Geographic capture loads a public page through a supported country, checks the actual exit location, and keeps the screenshot and other evidence with that verification. Use it from the dashboard, the API, and MCP. Free workspaces can still capture without a country.
Datacenter capture covers 18 supported countries, including Brazil and Germany. The live list, with names and ISO codes, is GET /capture/geo.
SiteLens checks the observed country through the same proxy before returning evidence. A mismatch fails closed.
Capture two or more countries in the dashboard and compare screenshots with each side’s verification metadata.
Automation. Observability. Diagnostics. Visual QA. Comparison. Reports—one product, not a screenshot tool or a DevTools wrapper alone.
Flows, scenarios, profiles — discoverable, editable, runnable across Desktop, CLI, and MCP.
Console, page errors, network exchanges—structured for agents.
Deterministic findings with evidence refs and recommended actions.
Labeled screenshots correlated with browser behavior.
Passing vs failing runs; new problems and timing regressions.
Portable evidence packages for humans and AI agents.
SiteLens automation is not mystery JSON buried in a project. Desktop turns local flows and scenarios into something humans can discover, understand, author, run, and debug visually — while the same files stay portable for CLI and agents.
When a step fails, open the browser evidence already captured for that run — console, network, screenshots — through Browser Intelligence. Don’t just know that the flow failed. Understand what happened in the browser.
Discover the flows and scenarios already in your project. See execution order visually, understand profiles and parameters, edit properties, run them, and inspect exactly where they failed.
Desktop Automation →Record a real browser session with Playwright Codegen. SiteLens converts it into a native flow you can review, edit, run, reuse, and expose to agents — from Desktop, CLI, or MCP.
Record Flow →One run. One evidence model. One diagnostics engine. One timeline. One report.
Automation and capture share a single Chromium execution.
Screenshots, console, network, payloads, bodies, diagnostics, timeline, traces, videos, reports—correlated.
Navigate the entire browser execution in order—steps, events, findings, artifacts.
SiteLens correlates screenshots, console, network, request payloads, response bodies, diagnostics, timeline, traces, videos, and reports into one coherent evidence model—so agents retrieve a chain, not a folder dump.
Compare a passing run with a failing run; identify new browser problems; detect timing regressions; correlate visual changes with browser behavior.
Learn moreFlagship navigation of an entire browser execution—steps, console, network, diagnostics, and artifacts in one correlated timeline.
Learn moreInvestigate a failed login. Inspect a network request. Explain a failed assertion. Compare successful and failed runs. Diagnose a visual regression. Help create automation from recorded browser behavior, then retrieve the evidence chain—through SiteLens MCP as the preferred browser workflow platform.
Visual Automation workspace: explore flows/scenarios/profiles, Record Flow via Playwright Codegen, run with result overlays, and inspect Browser Intelligence evidence.
Download Desktop 0.2.2 →Cursor and other hosts—run flows, start Codegen sessions, convert recordings, and inspect evidence. Local free; Cloud when you add a key.
Codegen tools →Local QA, sitelens flow codegen, reports, and doctor checks from the terminal.
SiteLens Cloud/API for CI and persisted runs.
How you run Browser Intelligence — Local, Cloud/API, and Studio on one engine (plus Theme for design intelligence).
Free local browser intelligence on your machine—localhost, staging, and private URLs.
Hosted execution, persisted runs, API-triggered workflows, and CI.
Inspect runs, templates, findings, and evidence—not the whole platform.
SiteLens Local is free local execution on your machine—not a limited SaaS tier. SiteLens supports Chromium, Firefox, WebKit, and installed Chrome/Edge channels through Playwright against localhost, staging, and private URLs you can reach from your workstation. No API key required; reuse Playwright-style auth state where supported. Prefer a native macOS app? Download SiteLens Desktop for macOS (version 0.2.2).
Playwright-powered Chromium, Firefox, and WebKit (plus Chrome/Edge channels) with screenshots and step reports—cloud features are separate; local execution stays free.
QA apps before they are public: dev servers, VPN staging, and LAN hosts your browser can open.
Run from terminal, scripts, or SiteLens MCP in Cursor—same Local engine from your IDE.
Same browser QA semantics—different runtime. SiteLens Local is free on your machine; SiteLens Cloud/API is for hosted runs, persistence, and operational workflows.
Scenario Runner chains reusable flows in a shared browser session via SiteLens Local and MCP—saved auth profiles, parameterized flows, and multi-step workflow verification for authenticated apps.
Two capabilities that extend SiteLens beyond smoke tests: structured page understanding for agents and IDE-native verification gates that run real browser flows before you ship.
Render a page once and get ranked primaryActions, secondaryActions, navigationActions, and inferred workflowStages—confidence-scored summaries agents can plan against without scraping raw DOM.
REST POST /analyze/page · pairs with Theme for design + intent · feeds agent verification planning.
Page Intelligence details
Keep Cursor (and other MCP hosts) honest: run repo-local flows via sitelens_run_local_flow, compare screenshots, and use prompt-library gates so agents verify in Chromium—not by claiming “looks good.”
Free local execution · project flows under .sitelens/flows · optional Cloud/API when you add a key.
Start from proven flow templates—not blank scripts. Each run produces structured pass/fail summaries, dedicated result panels, and named screenshots you can share with teammates or wire into CI.
Check internal links without writing a crawler.
Exercise a bounded set of same-site links after deploys or nav changes—see which pages load, capture lightweight screenshots, and get a clear broken-link rollup in QA Studio.
Verify your mobile menu opens and exposes navigation.
On a phone-sized viewport, find a likely menu control, open it, and confirm nav content appears—with before/after screenshots and plain-language failures when the toggle or drawer is missing.
Confirm a deploy loads on desktop and mobile.
Post-deploy smoke on desktop and mobile viewports: page load, visible body, heading or main landmark, stable screenshot slugs, and CI-friendly pass/fail—without clicking through your whole app.
Most browser QA tools stop at a page. Real applications require setup. SiteLens can reuse saved browser profiles, initialize application state, run setup flows, and verify the workflow your users actually follow.
Chain reusable flows together using one browser session.
Reuse a saved auth profile or run a login setup flow.
Select test data, toggle features, or seed application state.
Navigate to the route or feature under test.
Click, fill, and interact across multi-step journeys.
Assert UI state, capture screenshots, and export artifacts.
Pass IDs, URLs, environment values, and test data into reusable scenarios without rewriting flows. Scenario JSON under .sitelens/scenarios/ accepts params and {{placeholders}} that flow into each chained step.
Cookies, local storage, session state, and authentication persist across the workflow. Login once, run setup flows, then verify—without restarting the browser between steps.
Run scenarios via SiteLens MCP (sitelens_run_local_scenario) or the CLI. See the Scenario Runner guide and QA Studio scenarios.
SiteLens combines browser automation, observability, diagnostics, visual regression, and evidence into one coherent workflow—Local, Cloud/API, Studio, Reports, and MCP.
The umbrella platform: automation, observability, diagnostics, comparison, timeline, and evidence.
The QA engine behind SiteLens Local and SiteLens Cloud/API—real browser execution, workflow verification, and structured evidence.
Evidence-first reporting — turn run artifacts into portable JSON, Markdown, and offline HTML without uploading evidence.
Template runs, screenshot review, structured findings, and flow inspection on the shared QA engine (not a separate runtime).
Typography, colors, and confidence-scored tokens from rendered pages—for embeds and match-my-site flows.
Integration layer for SiteLens Local in Cursor — install @sitelensapi/mcp from npm; optional Cloud/API tools when you add SITELENS_API_KEY.
Live samples from SiteLens Cloud/API when the API persists runs (SiteLens Local runs stay on your machine). Select a row for screenshots and step detail.
Tell us what you are building or what you want SiteLens to validate — localhost apps, staging URLs, deployment checks, MCP workflows, Theme API usage, or Cloud/API runs.