Summary
Capturing the element, not just the complaint, is what makes browser feedback actionable.
A screenshot tells the agent what is wrong; the captured element tells it where. Without that, every piece of feedback starts with the agent guessing which component you meant. Keeping a history matters for the same reason it matters in any review loop — the second round needs to know what the first one said. The author credits their inspirations openly, which is a small good sign about a small project.
What it is
A pairing of a server and a Chrome extension so a person can leave feedback in the browser and an agent can read it, with the two connected over a live socket.
What you get
- Interactive feedback collection through the extension interface
- Text and image attachments
- Real-time communication rather than polling
- Feedback history, saved and managed automatically
- Element capture and screenshots from the page
Requirements
The Chrome extension alongside the server. MIT licensed. Documentation is in Chinese.
Setup effort
One command — npm install -g mcp-chrome-feedback@latest