FAQ

FAQ

What is Loupe?

An open-source visual-feedback SDK. A developer embeds it in a product; authorized users (PMs, QA, designers) get an in-page toolbar to inspect any element, pin a comment, and capture a screenshot. Feedback is stored and can be read by Claude Code via MCP.

How do comments survive a redeploy?

Each comment stores a multi-signal fingerprint of its target element and re-resolves it by scoring candidates on load and on DOM changes. See Re-anchoring.

How does Loupe connect to Claude Code?

Through an MCP server exposing list_comments, get_comment, and update_status. Claude reads the backlog with full element context and resolves it.

Does the widget affect my site's styles or performance?

No. It renders in a Shadow DOM (no CSS leakage either way), loads asynchronously, and only activates for authorized users.

How is Loupe different from BugHerd / Marker.io / Userback?

It's open source and closes the loop to code — instead of stopping at an issue tracker, it hands comments to an AI agent via MCP with full context.

Is it free?

Yes — MIT licensed. Source on GitHub.

Can I use my own Postgres / object storage?

Yes. Set DATABASE_URL for hosted Postgres; the blobs.ts seam swaps to S3/R2. See Architecture.

Which browsers are supported?

Modern evergreen browsers for the SDK. The extension targets Chromium (MV3).