Disclaimer
Is this project open source? Yes
Is this project completely free? Yes
Is this project vibe-coded beyond the author’s ability to comprehend how it works? Yes
Community Directory: community.obsidian.md/plugins/promptuary
Hi everyone,
I’m building Promptuary, an Obsidian plugin for people who read, review, and revise a lot of Markdown docs.
The problem I kept running into was this:
When I read a PRD, technical design, draft article, or research note, I often know exactly what should be changed. But my comments end up scattered between highlights, side notes, and a separate AI chat. By the time I ask an AI tool to revise the document, I have to explain everything again.
Promptuary tries to make that workflow smoother:
- Highlight and take notes while reading
- Leave natural-language review comments directly on selected text
- Keep annotations in sidecar JSON, so the original Markdown is not polluted
- Export all review comments as an AI-ready instruction file
- Copy a full prompt with the original text + review comments
- On desktop, send the task to Claude Code, Codex CLI, Aider, Gemini CLI, or a direct API provider
- Preview the AI-generated changes as a diff before accepting or rejecting them
The goal is not to let AI blindly rewrite your notes. The goal is to capture your review intent while you are reading, then let an AI agent do the mechanical editing work, with you still confirming the final diff.
Some use cases I built it for:
- Reviewing product docs / PRDs
- Editing article drafts
- Giving feedback on technical documents
- Cleaning up research notes
- Practicing interview answers and revising them section by section
I have currently verified Claude Code on macOS and direct API calls; other invocation methods are still being tested, as they are subject to subscription requirements.
I’d love feedback from Obsidian users who review long notes or documents regularly. In particular, I’m curious whether your review workflow usually starts from highlights, comments, or a separate AI chat.