Skip to content

Ship what
you designed

Pin feedback on live sites and Figma, let layernote measure every difference between design and production, and hand the fixes to your coding agent.

terminal
claude mcp add layernote --env LAYERNOTE_TOKEN=mcp_xxx \  -- npx -y @layernote/mcp

Set up your coding agent

Paste this into Claude Code, Codex or Cursor inside your repository. It connects the layernote MCP server, works through the open feedback, and closes each item with a note on what changed.

Agent prompt

Help me connect this project to layernote. Read https://layernote.io/docs/ai-access first. Then: 1. Register the layernote MCP server with the token I give you: claude mcp add layernote --env LAYERNOTE_TOKEN=<token> -- npx -y @layernote/mcp 2. Call list_feedback and work through every open item: open it with get_feedback, fix it in this codebase, and explain what you changed with add_feedback_comment. 3. For comparison findings, finish with verify_feedback rather than resolve_feedback, so layernote checks the deployed page before anything is closed. 4. Mark false positives with wont_fix_feedback and a clear reason. Ask me before changing anything outside this repository.

Guides

Something missing?

These docs grow with the questions we get. If a page does not answer yours, write to info@layernote.io.