TRUST Review 作者： Samuel Mok

TRUST Review helps researchers and educators conduct structured evaluations of information tools — search engines, academic databases, and AI assistants — using the TRUST Framework developed at the University of Twente.

Everything happens inside a browser sidebar:

• Capture evidence — screenshot the active tab at native resolution and archive the page HTML.
• Annotate — draw arrows, boxes, labels, and freehand strokes over captures with a built-in canvas editor.
• Score — evaluate each rubric principle on a 0–3 scale (with N/A and "unsure"), attaching the captured evidence directly to each item.
• Export — finish with a self-contained review package: CSVs of scores/metadata plus a standalone HTML report (fully inlined, viewable anywhere).

Multi-session support keeps every review independent, with debounced auto-save to local storage. Switch between reviews, resume incomplete ones, or start fresh.

Privacy. The extension is fully local. There is no backend, no account system, and no analytics or telemetry — all sessions, screenshots, and exports live in the browser's IndexedDB and never leave the machine. The only outbound connection is to a vendor CDN (cdn.tldraw.com) that serves UI translations for the annotation canvas; no user data is sent there.

Built with WXT and React. Source code is public under the Apache-2.0 license at https://github.com/utsmok/review-extension.