| .github | ||
| src | ||
| .gitignore | ||
| CHANGELOG.md | ||
| esbuild.config.mjs | ||
| eslint.config.mjs | ||
| LICENSE | ||
| manifest.json | ||
| package-lock.json | ||
| package.json | ||
| README.md | ||
| styles.css | ||
| tsconfig.json | ||
| versions.json | ||
| vitest.config.ts | ||
Readability Compass
Write at the level your readers need — and know it while you write, not after they've given up.
You know the feeling: you wrote a careful explanation for a broad audience, reread it twice — and you still can't say whether the people you wrote it for will actually follow it. Word counters tell you how much you wrote, never how hard it reads. The readability tools that do exist speak English-only formulas that mean nothing in your language, or dump thirty raw numbers on you without a verdict. So you guess. And your readers pay for the guess.
Readability Compass replaces the guess with a signal, right where you write. It measures your note with LIX — a readability score built only on sentence length and the share of long words, so it needs no syllable guessing or word lists and works in Dutch, English, German, the Scandinavian languages and any language that separates words with spaces. Then it holds that score against the target band you choose (≈ CEFR B1, B2, C1, or your own ceiling) and gives you the one verdict that matters, permanently, in the status bar: on target, or not.
(Why "Compass"? A compass doesn't walk for you — it tells you, at a glance, whether you're still heading where you meant to go. Your writing stays yours; the plugin only keeps you honest about the direction.)
Get started in 60 seconds
- Install & enable: Settings → Community plugins → search for Readability Compass.
- Pick your audience in the settings: ≈ B1, B2, C1, or a custom LIX ceiling.
- Write. The status bar keeps score:
LIX 42 · ≈ B2 · ✓ · 351 w. Click it whenever you want the why.
That's the whole entry fee. No account, no API key, nothing leaves your machine.
What you get
- Status bar, always on — score, band, target check and word count, with configurable segments. Select text and it scores the selection. Click it to open the panel.
- Side panel — the why behind the verdict: a score card (LIX, band, indicative CEFR audience, target check), the published Flesch variant for your language (ten supported), counts (words, sentences, words per sentence, long words, paragraphs, reading time), your hardest paragraphs (by LIX) and your longest sentences — click one and you jump straight to the culprit. You choose how many entries the lists show, reveal more with Show more, and can hide sentences below a word threshold.
- Score a folder or a selection of notes — select notes in the file explorer and the panel scores them together, live; change the selection and the score follows (right-click menu for folders and larger sets). One combined score, a per-note list sorted hardest-first (each note against its own target), and the longest sentences across all of them — click to open that note at that sentence.
- Five commands — open the panel · score the current note · score the selection · insert a readability report at the cursor (as a callout) · toggle the status bar.
What Readability Compass is not
A compass, not a co-author:
- Not a grammar or spell checker. It measures and points; it never rewrites.
- Not an AI writing assistant. No suggestions, no rephrasing — the writing stays yours.
- Not a statistics dashboard. Vault-wide word counts and writing streaks are other plugins' turf; this one answers a single question — can your readers follow you?
Why you can trust the score
-
Only running text counts. Front matter, code blocks, inline code, tables, link URLs, images, HTML and tags are stripped before measuring — without losing positions, so jump-to-sentence lands exactly. Notes that are tables (question banks, glossaries) can opt in: turn on Include table text, or set
readability-tables: truein a note's front matter. A manual selection always measures everything you selected — tables and code included. -
One published, inspectable formula:
LIX = words / sentences + 100 × long_words / words (long word: > 6 letters)LIX Band Indicative audience < 30 very easy ≈ B1 30–40 easy ≈ B1 40–50 average ≈ B2 (≤ 45) / ≈ C1 50–60 difficult ≈ C1 (≤ 55) / ≈ C2 > 60 very difficult ≈ C2 -
Honest labels. The CEFR labels are an indication of the audience a text suits, not a certified level. Scores are hidden below a minimum word count (default 40) because they are too erratic on short texts.
-
A second opinion where one exists. When the language is known (auto-detected, or fixed in the settings), the published Flesch variant is shown alongside: Flesch Reading Ease (en), Flesch-Douma (nl), Flesch-Amstad (de), Fernández-Huerta (es), Kandel-Moles (fr), Flesch-Martins (pt), Flesch-Vacca (it), Flesch-Oborneva (ru), Ateşman (tr) and Flesch-Bendová (cs). Their syllable counts are heuristic (±1 on tricky clusters); LIX stays the main score. Missing your language? Open a language request — adding one is a single entry in the language registry (a stopword set, a syllable counter and published formula coefficients). LIX itself already works for your language today if it separates words with spaces.
-
Known limitations, stated. Sentence splitting is punctuation- and line-based (abbreviations like "e.g." count as sentence ends), and LIX assumes space-separated words — it is not meaningful for CJK text. Keyword-style notes (slide decks, bullet lists of long content words without connecting prose) score deceptively high: every line counts as a sentence and the long-word share dominates. LIX is meant for running prose — turn on Mark long words to see exactly which words drive such a score.
Privacy — what leaves your machine
Nothing. Every score is computed locally: no network calls, no telemetry, no account.
Install
Community plugins: Settings → Community plugins → search for Readability Compass → Install → Enable. Or start from the directory listing.
Manual: download main.js, manifest.json and styles.css from the latest
release and copy them into
<vault>/.obsidian/plugins/readability-compass/.
Works on desktop and mobile (Obsidian 1.7.2+).
Works well with
Voxtral Transcribe — from the same workshop. Spoken first drafts are honest but rambling: speak your idea on a walk, let Voxtral turn it into a markdown note, then let Readability Compass show how far that transcript is from your audience — your longest sentences are one click away. Capture by voice, tighten by compass.
Parallax — also from the same workshop. Parallax guards the thinking behind your conclusions (graded evidence, belief tracking, a methodological account); Readability Compass guards whether the write-up of those conclusions reads at the level your audience needs. Sound research on one side, readable reporting on the other.
The three plugins share the same principles — your notes stay yours, local where possible, no telemetry — but stay deliberately separate tools, each owning its own job.
Support
Readability Compass is free and GPL-3.0. If it keeps your writing on target, you can buy me a coffee ☕ — it keeps the work going.
Issues and ideas: GitHub issues. Development happens in the codebase-basecamp repo; this repo mirrors the source and builds each release here, so every release asset carries a build-provenance attestation you can verify.
License
GPL-3.0 — © 2026 Max Kloosterman