Contextual feedback capture
Leave written notes, screenshots, and voice critiques on the exact thing you are reviewing instead of describing it abstractly in chat.
Remarc
Remarc is a macOS feedback layer for AI collaboration that lets you add comments, screenshots, and voice critiques to on-screen context and send that structure to supported agents through MCP. It is free and open source under the MIT license.
Overview
Remarc is a contextual feedback layer for AI agents. It sits in your Mac menu bar and is designed to capture the “last 10%” of AI-assisted work: the review, QA, and iteration that happen after a draft or prototype is already close.
Instead of typing a paraphrase into chat, Remarc lets you annotate what is actually on screen with comments, screenshots, and voice critiques. The captured context is then available to your agent through MCP, so the agent can work from the original object rather than a description of it.
Core features
Structured screen context
Attach the selected text, source app, screenshot region, or web element so the agent receives the specific artifact you were looking at.
Browser-page context capture
Use the Chrome extension to capture web context, including page URL, selected text, surrounding content, element data, styles, and structure.
Local-first storage
Keep feedback in a local file on your Mac with no accounts or cloud backend for comments, so the information stays local unless you send it through an integration.
Agent integration path
Connect through MCP so supported agents can read comments, work through them, and mark items resolved; the site specifically mentions Claude Code, Codex, and Cursor support.
Practical use cases
Final-pass QA on AI work
Review a near-finished AI-generated design or implementation and point directly at the visual issue, missing detail, or layout problem instead of writing a long explanation.
Web UI review in the browser
Attach comments to specific web elements, selected text, or page regions while testing a browser-based experience, so the agent can see exactly what changed and where.
Fast verbal feedback capture
Record a quick voice critique when typing would be slower, then let the agent receive that feedback in the same contextual package as the screen artifact.
Agent review loops
Leave structured notes that a supported agent can revisit and resolve later, which is useful when iterating in multiple passes rather than fixing everything in one conversation.
Pros and Cons
Pros
- Provides comments, screenshots, and voice critiques tied to on-screen context rather than a text paraphrase.
- Uses MCP to pass structured feedback to supported agents.
- Supports direct browser context capture through the Chrome extension, including element and region capture.
- Stores comments locally on the Mac and states that it has no accounts, telemetry, or cloud backend for comments.
- Free and open source under the MIT license, with no subscriptions or paid tiers stated on the site.
Cons
- The Chrome extension is not on the Chrome Web Store yet, so installation is manual.
- The source material does not show a public pricing page beyond a 404 at /pricing, so there is no separate pricing detail to review here.
- Source coverage is limited on broader platform support, team features, and non-browser workflows beyond the documented context-capture paths.
FAQ
What does Remarc do?
Remarc is a contextual feedback layer for AI agents. It lives in your menu bar and lets you leave comments, screenshots, and voice critiques on what you see on screen, then passes that context to supported agents through MCP.
How much does Remarc cost?
Remarc is free and open source under the MIT license. The site says there are no accounts, subscriptions, or paid tiers.
Which AI agents work with Remarc?
The homepage says Claude Code and Codex connect through the Remarc plugin, while Cursor is set up directly from the app’s preferences. It also notes that any agent with MCP support can read comments, work through them, and mark them resolved.
Is my data private?
Remarc stores comments in a local file on your Mac and says it has no accounts, telemetry, or cloud backend for comments. Voice transcription runs on-device, and data leaves only through agents or integrations you explicitly use.
How do I install the Chrome extension?
The Remarc Chrome extension is installed manually from the download page rather than the Chrome Web Store. The setup guide walks through unzipping the download, loading it as an unpacked extension, enabling it in Remarc settings, and reloading the browser tabs you want to capture from.
Quick Facts
- Category
- AI feedback layer
- Platform
- macOS 14.0 or later
- License
- MIT
- Distribution
- Free and open source
- Primary workflow
- Screen-based comments, screenshots, and voice critiques for AI agents
- Source domain
- remarc.app
Alternativas ao Remarc
Biji
Biji é uma plataforma versátil projetada para aumentar a produtividade por meio de ferramentas e recursos inovadores.
Dcyde
Dcyde is a decision log for product teams that captures what was decided, the reasoning behind it, and the source in one shared place. It supports pinning decisions from Slack, Figma, or inside Dcyde and keeps the record searchable over time.
Carbon Voice
Carbon Voice is an asynchronous voice messaging app for teams and individuals, with transcripts, AI catch-up, and cross-device access. It helps people and agents communicate without needing a live call.
Goldfish
Goldfish is an AI memory app for macOS that helps you reply, write, summarize, and continue using the context already on your Mac. It stores memory locally and writes in your tone.
ArtDeck
ArtDeck is a visual reference board app for iPhone, iPad, and Mac that lets artists collect mixed reference material on one canvas and study it with ToolBox lenses. It works offline, supports optional iCloud sync, and is sold as a one-time purchase with no account required.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.