A Build Kit to help you build your own version of a Markdown Notes App using AI coding agents like Claude Code, Codex, or Gemini.
What you'll build
A capture-first note-taking web app that turns quick thoughts and uploaded files into cleanly tagged markdown saved as plain files in the user's own cloud storage. Structured tags drive automatic folder routing, so notes self-file as they're captured.
The app is single-user, mobile-friendly, and works on a phone. No proprietary database for note content — every note is a real markdown file the user owns and can take anywhere.
Who it's for
The product you'll build is for solo knowledge workers who want their notes owned as plain files in their own cloud storage account, and who hate manually filing every new note. Especially:
- Writers, researchers, builders, and creators who want a clean markdown vault they fully control.
- People who think in tags — they want to label every note's project, type, or status at the moment of capture.
- Anyone who hates manual filing and wants notes to land in the right folder automatically.
What it solves
- Note-app lock-in. Notes live as plain markdown files in the user's own cloud storage, not in a proprietary database.
- The "where do I put this?" tax. Capture-time tags decide where each file lands automatically, based on rules the user sets once.
- Tagging as an afterthought. Tags are first-class buttons in the capture flow — large, tappable, designed for thumbs.
- Markdown without metadata. User-defined front matter fields enforce consistent, machine-readable tags on every note.
- Hard to hand notes to an AI agent. Because every note is a real markdown file on the user's file system, the user can share a file path or a direct link and an AI coding agent can read or edit the note in place — no exports, no copy-paste.
Key user flows
- A user signs up, signs in, and connects their cloud storage account.
- A user types a note, taps a few tag buttons, and saves — the note lands in the right folder with the right filename automatically.
- A user drags a file (PDF, image, anything) into the upload tab, tags it, and saves — same auto-routing.
- A user opens any markdown file from the file browser and edits it in a clean markdown editor with auto-save.
- A user shares a file path or direct link to a note with an AI coding agent, which reads or edits the markdown in place.
- A user finds notes later through recents, pinned favorites, or full-content search.
- A user defines fields, options, and routing rules once in settings — captures self-file from then on.
How you'll build it
The build is broken into 5 milestones. Each one is a self-contained working session for an AI coding agent — you can stop, run what you've built, and pick up cleanly on the next one.
- Project scaffold, auth, and theming — stand up the project, add email/password auth, and wire in the visual language.
- Connect cloud storage and browse files — connect the user's cloud storage account via OAuth and let them navigate folders.
- Capture, save, and edit markdown notes — close the round-trip: capture, save to storage, open back, edit with auto-save.
- Structured tagging with auto-routing — the differentiator: define fields and rules, watch notes self-file by their tags.
- Find notes again — recents, drag-reorderable favorites, and full-content search across the vault.
What's included
When you download the kit, you get a README.md and a _build_plan/ folder you drop into your project:
README.md— orientation and how to use the kit_build_plan/PRD.md— the product requirements, data model, and milestone breakdown_build_plan/tech-stack.md— customizeable tech stack for your codebase_build_plan/design-system.html— visual reference (colors, type, components — open in a browser)_build_plan/milestones/<N>-<slug>/prompt.md— one folder per milestone with a lean, ready-to-paste prompt that works with any AI coding agent
