text-editor

To edit annotated text, submit content containing %%...%% instructions and ==...== highlights. The system applies explicit directives exactly, rewrites weak passages using context, strips all markers, and outputs a single clean block of text without commentary.

In AI editing annotations, %%instructions%% are explicit directives applied exactly as written to surrounding text, while ==highlights== flag weak or vague passages for Claude to rewrite using contextual judgment and inferred intent.

To apply inline comments and polish a draft, embed %%...%% directives for exact edits and ==...== markers around vague passages. The system processes both annotation types, removes the markers, and returns clean edited text only.

Yes, you can clean up annotated transcripts while preserving non-annotated content. The system detects and removes all %%...%% and ==...== markers, applies the requested edits, and outputs text with original structure intact for unannotated sections.

The best way to apply edits to AI-annotated notes is to use inline %%...%% and ==...== markers. This approach separates explicit instructions from contextual rewrite flags, ensuring clean output without manual marker removal.

No, the edited text output contains no commentary or formatting explanations. The system strips all annotation markers and returns only the single clean block of edited text, preserving structure while removing every %%...%% and ==...== marker.