Does this work with Claude Code?

Yes. Claude Code can call the Viewpo HTTP API directly using fetch or curl, parse the JSON response, fix any issues found, and re-audit — all within a single conversation.

How much does API access cost?

The free tier includes 50 API calls per month. Pro ($19/mo) gives you 500 calls with full audit features. Team ($49/mo) offers 2,000 calls with unlimited batch processing.

Close the visual feedback loop for AI-generated code.

Your AI builds the frontend. Viewpo's API tells it whether the result is actually responsive — in structured JSON it can act on.

The problems you know too well

AI can't see what it builds

Claude, GPT, and Cursor generate CSS but can't see the rendered output. There's no visual feedback loop — you're the eyes.

No visual quality gate

Lighthouse catches performance regressions. ESLint catches code style. Nothing catches "the hero overflows at 375px" — until a real user reports it.

Manual review is the bottleneck

Your AI agent can write and deploy code in seconds. But it still needs you to resize the browser and check if it looks right. That's the slowest step.

How Viewpo helps

The features that matter most for your workflow.

HTTP API with structured JSON

Send a URL, get back screenshots, per-check audit results, and an A-F grade as structured JSON. Your agent can parse it and act on it immediately.

Automated re-audit loop

Agent calls API → gets Grade D with 7 issues → fixes the CSS → re-audits → gets Grade A. No human in the loop. Reports success with the report link as proof.

CI/CD integration

Add a Viewpo audit step to your pipeline. Every preview deploy gets checked automatically. Grade below B? The check fails before merge.

Works with any AI coding tool

Claude Code, Cursor, Windsurf, Copilot, Bolt, Lovable — any tool that can make HTTP requests can use the Viewpo API to close the visual feedback loop.

Interactive sessions

Open a persistent browser session. Your AI clicks elements, fills forms, and scrolls through pages using compact element refs — a fraction of the context of raw DOM. Stream the session live via WebSocket.

Your Claude Code agent just built a landing page. It looks great in the code. But does it work on phones? The agent calls the Viewpo API, gets Grade D — 7 overflow issues. It fixes the CSS, re-audits, gets Grade A. It posts the report link in the PR comment as proof. No human ever resized a browser.

Frequently Asked Questions

Quick answers to help you decide.

The API returns JSON with: overall grade (A-F), per-viewport screenshots (base64 or URL), per-check results (pass/fail with details), and metadata (URL audited, timestamps, viewport sizes used).

Read all FAQs

See How Viewpo Compares

Already using another tool? See how Viewpo stacks up.

Viewpo vs Google Lighthouse → Viewpo vs BrowserStack → All comparisons →

See how your site looks on every screen

Paste any URL and get a responsive design audit with real browser screenshots and an A-F grade. No install required.

Free during beta. Plans from $19/mo at launch — early supporters get a lifetime discount.