ByteZone
Back to listClaim this project

metedataDeveloper tools

Remarc

A macOS feedback layer that captures screen context and comments for AI agents via MCP.

  • Design
  • QA / Testing
  • Collaboration
  • Desktop
  • Browser extension
  • macOS
  • Browser
  • Self-hostable
  • Runs locally
Open websiteView on GitHub
Remarc screenshot
Popularity
66 Stars
GitHub stars
Recent activity
9/9/2026
Updated in the last 30 days
License
MIT
Permissive

Why it matters

We look beyond stars: what problem it solves, whether it creates real utility, and what makes its approach worth noticing.

Last 90 days

Problem

Solves the friction of conveying visual, textual, and web context to AI coding agents. Traditional methods require manual screenshotting, copying text, and writing verbose prompts, which often leads to context loss.

Practical value

Highly practical. Running from the macOS menu bar, it supports text selection, screenshot annotations, web element capture (with CSS and React metadata via an extension), and local voice transcription. All data remains local and connects to agents via MCP.

Innovation / differentiation

Innovates by consolidating OS-level multi-modal captures (screenshots, voice, text) and web DOM context into a standardized MCP server, allowing AI agents to query structured feedback directly rather than relying on manual descriptions.

Leverage potential

Leverages the open MCP standard to connect with major clients like Claude Desktop, Claude Code, and Cursor without building proprietary integrations for each. It also supports webhooks and Markdown/JSON exports for workflow automation.

Why now

Excellent timing. As AI coding agents evolve from simple chat interfaces to autonomous workspace assistants, and with the rapid adoption of the MCP standard, there is a strong demand for tools that bridge local context and AI agents.

Community activity

In the last 90 days there were 11 new issues and 8 pull requests; the bounded issue/PR samples include 4 issue authors and 3 PR contributors, with 10 releases.

Maintainer responsiveness

The 11-issue window sample had a 100% close rate, and the 8-pull-request sample had a 75% merge rate. Maintainer-response observations covered 100% of that issue sample, with a 82% response rate and median first response of 8.4 hours.

3 contributors in PR sample10 releasesIssue response rate 82% · sample 11 (100% coverage)Median first response 8.4hPR merge rate 75% · 8 sampled in window

Key highlights

  • Multimodal Feedback Capture: Comment on selected text, annotated screenshots, web elements (with CSS, DOM, and React data), or record voice notes.
  • Local-First Privacy: All data stays on your Mac with no accounts or telemetry, only shared when you connect to an agent or trigger webhooks.
  • MCP Native Integration: Connects with Claude Desktop, Claude Code, Cursor, and other agents using the Model Context Protocol.

Quick start

How it is installed, how hard it is, and where to start.

Where it runs

To be confirmed

Difficulty

To be confirmed

Self-hostableRuns locally
  1. 01Download and install the Remarc client from the official website or GitHub Releases.
  2. 02Install the companion Chrome extension to capture web elements and CSS context.
  3. 03Select text, capture a screen region with annotations, or record voice feedback on your Mac.
  4. 04Organize the captured feedback cards into specific sessions within Remarc.
  5. 05Configure and connect your AI agent (such as Cursor, Claude Code, or Claude Desktop) to read and resolve comments via MCP.

Best for

  • Teams that want data on their own servers
  • Developers who want to try it on their machine

More about it

Remarc is an open-source feedback layer designed for macOS to streamline collaboration with AI coding agents. It allows you to point at anything on your screen—text, screenshots, web elements, or voice—and leave comments. Remarc automatically captures the rich underlying context (such as source apps, web CSS, accessibility data, or React components) and exposes it to AI agents via the Model Context Protocol (MCP), enabling agents to read and resolve your feedback directly without manual prompt building.

Sources

Each field shows its status and source — expand to review.

14 · Expand

  • capability tags

    Verified

    collaboration

    Source: admin_cms · cms editor · 9/8/2026

  • editor note

    Verified

    {"en":"Remarc addresses the context-sharing bottleneck in human-AI collaboration. By capturing precise interactive elements (like CSS, DOM, selected text, or annotated screenshots) on macOS and browsers, and exposing them to AI agents via MCP, it allows agents to understand your feedback exactly as a human colleague would.","zh":"Remarc 解决了人与 AI 协作中的“上下文传递”痛点。通过在 macOS 系统底层及浏览器中捕获精确的交互元素（如 CSS、DOM、选中文本或带标注的截图），并利用 MCP 协议直接提供给 AI 智能体，它让 AI 能够像人类同事一样理解你的修改意见。"}

    Source: admin_cms · cms editor · 9/8/2026

  • how to use

    Verified

    {"steps":[{"en":"Download and install the Remarc client from the official website or GitHub Releases.","zh":"从官网或 GitHub Releases 下载并安装 Remarc 客户端。"},{"en":"Install the companion Chrome extension to capture web elements and CSS context.","zh":"安装配套的 Chrome 浏览器扩展程序，以便捕获网页元素和 CSS 上下文。"},{"en":"Select text, capture a screen region with annotations, or record voice feedback on your Mac.","zh":"在 macOS 屏幕上选择任意文本、截取区域并添加标注，或使用语音录制反馈。"},{"en":"Organize the captured feedback cards into specific sessions within Remarc.","zh":"在 Remarc 中将捕获的反馈整理到特定的会话（Session）中。"},{"en":"Configure and connect your AI agent (such as Cursor, Claude Code, or Claude Desktop) to read and resolve comments via MCP.","zh":"配置并连接你的 AI 智能体（如 Cursor、Claude Code 或 Claude Desktop），通过 MCP 协议读取这些反馈并开始自动修复。"}],"installAt":"unknown","difficulty":"unknown"}

    Source: admin_cms · cms editor · 9/8/2026

  • intro

    Verified

    {"en":"Remarc is an open-source feedback layer designed for macOS to streamline collaboration with AI coding agents. It allows you to point at anything on your screen—text, screenshots, web elements, or voice—and leave comments. Remarc automatically captures the rich underlying context (such as source apps, web CSS, accessibility data, or React components) and exposes it to AI agents via the Model Context Protocol (MCP), enabling agents to read and resolve your feedback directly without manual prompt building.","zh":"Remarc 是一款专为 macOS 设计的开源 AI 协作反馈工具。它作为用户与 AI 编程智能体（如 Claude Code、Cursor 等）之间的桥梁，允许用户直接在屏幕上的任何地方（文本、截图、网页元素或语音）进行标注和评论。Remarc 会自动收集相关的原始上下文（如源应用、网页 CSS、无障碍数据等），并通过 Model Context Protocol (MCP) 协议将其传递给 AI 智能体，使智能体能够直接理解并解决这些反馈，无需用户手动复制粘贴或重新编写提示词。"}

    Source: admin_cms · cms editor · 9/8/2026

  • Latest release

    Verified

    v1.2.3

    Source: GitHub API · latest_release=v1.2.3 · 9/20/2026

  • License

    Verified

    MIT

    Source: GitHub API · license.spdx_id=MIT · 9/20/2026

  • needs api key

    Verified

    No

    Source: admin_cms · cms editor · 9/8/2026

  • One-liner

    Verified

    {"en":"A macOS feedback layer that captures screen context and comments for AI agents via MCP.","zh":"专为 AI 智能体设计的 macOS 屏幕反馈与上下文收集工具。"}

    Source: admin_cms · cms editor · 9/8/2026

  • Platforms

    Verified

    macos, browser

    Source: admin_cms · cms editor · 9/8/2026

  • Category hint

    Inferred from materials

    ai-apps

    Source: Project README · hint=ai-apps · 8/21/2026

  • product forms

    Verified

    desktop, browser_extension

    Source: admin_cms · cms editor · 9/8/2026

  • role tags

    Verified

    design, qa

    Source: admin_cms · cms editor · 9/8/2026

  • supports local

    Verified

    Yes

    Source: admin_cms · cms editor · 9/8/2026

  • supports self host

    Verified

    Yes

    Source: admin_cms · cms editor · 9/8/2026

Related projects

Other verified projects matched by category, capabilities, and intended roles.

openai-agents-python

A Python framework by OpenAI for building multi-agent workflows, supporting text, voice, and sandbox-based agent applications.

Same Developer tools categoryBoth support Collaboration

flipt

A self-hosted feature management platform that stores feature flags in Git and manages multi-environment delivery through a web UI, APIs, and real-time updates.

Same Developer tools categoryBoth support Collaboration

massing

Open-source, self-hosted, IFC-native AEC platform featuring in-browser BIM modeling, a general contractor portal, and development proforma tools.

Same Developer tools categoryBoth support Collaboration

spec-kit

An open-source CLI toolkit for Spec-Driven Development that guides AI coding agents through structured specification, planning, and code generation steps.

Same Developer tools categoryBoth fit QA / Testing

playwright

A cross-browser automation framework for testing and AI agent control, supporting Chromium, Firefox, and WebKit.

Same Developer tools categoryBoth fit QA / Testing

puppeteer

A JavaScript library that provides a high-level API to control Chrome or Firefox over DevTools Protocol or WebDriver BiDi.

Same Developer tools categoryBoth fit QA / Testing