AI Markdown Feedback
Code review for prose: annotate markdown with structured feedback that LLMs can read and act on.
LLM coding agents generate Markdown artifacts constantly - plans, docs, explanations. But when you want to give feedback on one, you’re stuck writing free-form text or making inline edits that lose context. There’s no structured way to say “discuss this”, “change that”, or “remove this section” in a format the LLM can parse.
Ace: AI Markdown Feedback is a VS Code extension that fixes that. It adds four annotation types to your Markdown files, all stored as plain-text syntax in the .md source - no sidecar files, no proprietary format:
| Syntax | Purpose |
|---|---|
==highlight this== |
Flag text for discussion |
%%your note here%% |
Leave a comment for the LLM |
> [!EDIT] Change X to Y |
Specific edit request |
~~remove this~~ |
Suggest deletion |
Any LLM can parse these annotations directly from the source file.
How you use it
Open any .md file, launch the feedback preview (side-by-side or preview-only), select text, and annotate with single keys - H to highlight, C to comment, E to suggest an edit, D to mark for deletion. Then hand the annotated file back to your LLM.
On the first annotation, the extension inserts an instruction header telling AI tools to treat the markers as feedback and not strip them. There’s also a “Copy AI Instructions” command that produces a ready-to-paste snippet for your CLAUDE.md, .cursorrules, or Copilot config, so your tools always know how to process the annotations.
Install
Search “Ace AI Markdown Feedback” in the VS Code Extensions sidebar, or install from the Marketplace.