Skip to content
← Back to work

web-remarq

2025 - Present·Creator & Maintainer
TypeScriptReactVueSolidnpm

Overview

An open-source library for placing visual annotations directly on web applications. Designers mark issues on staging, developers see them in code context, and AI agents receive structured export with source locations. Published on npm, supports React, Vue, and Solid.

Inside the product

Feedback that stays attached to the interface.

Select an element, leave a specific note and review it in place. This recording shows the real annotation tool running on its Nimbus demo page.

My roleCreator & maintainer

What I builtAnnotation UI / matching / developer tooling

16-second product recording · Nimbus is a sample interface used to demonstrate web-remarq.
01 / 03

Point at the actual interface element. The inspector identifies the target so feedback has a precise location.

1 / 3 · Inspect
web-remarq highlighting a dashboard metric and identifying its DOM element

Point at the actual interface element. The inspector identifies the target so feedback has a precise location.

Key Features

Framework-Agnostic Architecture

Problem

Design review tools are typically framework-specific or require heavy setup. Teams using different frameworks need a universal solution.

Solution

Built a zero-dependency core with framework adapters for React, Vue, and Solid. The core handles DOM fingerprinting, annotation state, and export - adapters are thin wrappers.

Result

5 npm packages, single API across all frameworks. Install and start annotating in under 3 minutes.

DOM Element Fingerprinting

Problem

Annotations need to survive page reloads and code changes. Traditional selectors (IDs, classes) break when developers modify markup.

Solution

Designed a 6-priority fingerprinting system that identifies DOM elements using multiple fallback strategies - from stable attributes down to structural position.

Result

Annotations reliably persist across page reloads and minor code changes without manual re-mapping.

AI Agent Integration

Problem

Design feedback typically lives in Figma comments or Slack - disconnected from code. AI coding agents can't access this context.

Solution

Built a structured JSON export with source locations. Annotations include component paths, element selectors, and designer notes - ready for AI agent consumption.

Result

AI agents receive design feedback with full code context, eliminating the manual translation step between design review and implementation.