Markdown Notes Index

About this tool

A folder of Markdown notes is only as useful as the note that points at the rest of them. This tool builds that note for you: paste your notes into the box one after another and it splits them apart, works out each note's title, collects its tags, reads its heading outline and counts its words, then renders the whole set as one linked index — the "map of content" a vault, a docs folder or a pile of meeting notes is usually missing.

Everything runs as WebAssembly inside this page. Your notes are never uploaded, and the output never reproduces a note's body — only its title, tags, outline and counts.

A worked example

Paste this bundle, leave every option at its default, and press nothing else:

# Getting started

Some intro text.

## Install

Run it.

# Weekly review

Notes about the week.

You get back:

# Notes index

2 notes · 0 tags · 14 words

## Contents

1. [Getting started](#getting-started)
2. [Weekly review](#weekly-review)

## Getting started

8 words · 1 heading

- Install

## Weekly review

6 words · 0 headings

The first # heading of each note became its title and its anchor link; the deeper ## Install heading stayed in that note's outline; the counts came from the note bodies.

Where notes are split

• Heading (default) — a new note starts at every top-level # heading. This is what a concatenated pile of notes usually looks like.
• Thematic break — a new note starts at every ---, *** or ___ line. A --- that opens a note's YAML front matter is not treated as a break, so front matter stays with its note.
• File markers — a new note starts at every === notes/todo.md === or ==> notes/todo.md <== banner, the shape head/tail print when they dump several files at once. Each note then remembers its path, so links point at the real files ([Today](notes/todo.md#today)) and a note with no heading falls back to its file name.

Titles, tags and links

A note's title is the first of these that exists: a title: key in its YAML front matter, its first ATX heading, its file name from a file marker, then Untitled note 3 as a last resort. Tags come from front-matter tags, tag or keywords — inline (tags: a, b), flow (tags: [a, b]) or block-list form — plus any #tag written in the body, which you can switch off. Duplicate titles get distinct anchors (#log, #log-1), so every link in the Contents list lands somewhere different.

Choose anchor links for a self-contained index note, wiki links ([[Note title]], [[Note title#Heading]]) for an Obsidian-style vault, or plain text for somewhere without link targets. Switch the output to JSON to feed the index into another script, or CSV to open it in a spreadsheet.

Limits and edge cases

• Headings must be ATX style (# through ######). Setext underlines (Title followed by =====) are not recognised as headings.
• Headings inside ``` or ~~~ code fences are skipped, so a # comment in a shell snippet never lands in your outline.
• Up to 500 notes per run. Over that you get a clear error rather than a silently truncated index — split the bundle and index it in batches.
• Front matter must be a --- fenced block at the very start of a note, and only title, tags/tag and keywords are read from it. Everything else is ignored, not an error.
• Inline tags need at least one letter, so issue references like #1234 stay out of your tag list, while #build/ci is kept whole.
• Words are whitespace-separated tokens containing at least one letter or digit, so bare Markdown punctuation (-, >, |) is not counted as a word.

FAQ