DEV Community: Qonspekt

Qonspekt now works with Gemini — free Obsidian notes, no credit card

Qonspekt — Tue, 12 May 2026 18:33:52 +0000

The biggest barrier to trying new AI tools? "I need to add a credit card."

Qonspekt just removed that barrier entirely. It now supports Gemini — and Google's free API tier means you can generate atomic Obsidian notes without spending a single cent.

What changed

Qonspekt now has a provider toggle: Claude or Gemini (FREE).

For Gemini, use Google's free API tier:

No credit card required
No billing setup
Gemini 2.0 Flash: free up to 15 requests per minute
Get your key in 30 seconds at aistudio.google.com/apikey

How to get started for free

Go to aistudio.google.com/apikey
Click "Create API key" — takes 30 seconds, no billing
Open Qonspekt
Click "Gemini FREE" → paste your key → generate notes

That's it. Zero cost.

What you get

Paste any article → Gemini extracts 3–7 key concepts → each becomes an atomic note with:

YAML frontmatter (title, tags, aliases)
[[wikilinks]] connecting related concepts
## Sources section
Download as ZIP → drag into Obsidian

Claude still works too

If you have Claude API credits, the Claude provider still works. Gemini is just the new free option that removes the "I need to pay first" friction.

Try it

→ qonspekt.github.io/qonspekt

Free Gemini key: aistudio.google.com/apikey (30 seconds, no billing)

Open source: github.com/Qonspekt/qonspekt

Web clippers vs. Qonspekt: what's the difference for Obsidian users?

Qonspekt — Tue, 12 May 2026 18:08:47 +0000

If you use Obsidian and want to save content from the web, you've probably tried a web clipper. Web2MD, WebInk, MarkDownload — they all do roughly the same thing: convert a webpage's HTML into Markdown.

That's useful. But it's not the same as taking notes.

What web clippers actually give you

You paste in an article URL, and you get back one Markdown file. The file is the article — its structure, its headings, its paragraphs — just in Markdown format instead of HTML.

What you don't get:

Identified concepts worth keeping
Atomic note structure
YAML frontmatter
Connections between concepts ([[wikilinks]])
Notes you can actually search and reuse

You still have to do the intellectual work: read the file, extract what matters, decide how to structure it, write atomic notes one by one.

What Qonspekt does instead

Qonspekt doesn't convert HTML. It reads the article and identifies what you'd actually want to keep in a Zettelkasten.

Input: a 3000-word article about the Feynman Technique.

Output:

feynman-technique.md — what it is, why it works
learning-by-teaching.md — the psychological mechanism
knowledge-gaps.md — how to find them
active-recall-vs-passive-reading.md — the contrast Feynman makes

Each note has YAML frontmatter, 2–4 tags, wikilinks connecting the concepts within the batch, and a ## Sources section. Download as ZIP, drag into Obsidian.

The key difference in one sentence

Web clippers preserve the article. Qonspekt writes notes about the article.

When to use which

Use a web clipper when:

You want to save a reference document for later
You need the full article structure preserved
You're archiving, not note-taking

Use Qonspekt when:

You want to add knowledge to your Zettelkasten
You're building an Obsidian vault of concepts
You read to learn, not just to collect

Practical comparison

	Web Clipper	Qonspekt
Output	1 Markdown file	3–7 atomic notes
Frontmatter	No	Yes (title, tags, aliases)
Wikilinks	No	Yes (between extracted concepts)
What it understands	HTML structure	Article concepts
Install required	Browser extension	No (browser-based, no install)
Cost per article	Free	~$0.003 (Claude API, BYOK)
Backend	None / server	None (browser-only)

Try Qonspekt

→ qonspekt.github.io/qonspekt

Bring your own Claude API key. No account, no backend. New Anthropic accounts get free credits.

If you've tried other approaches to this problem — Readwise, Notion web clipper, custom scripts — I'd be curious what actually stuck for you.

Using AI to build a Zettelkasten without the friction

Qonspekt — Tue, 12 May 2026 17:56:01 +0000

The hardest part of building a Zettelkasten isn't the theory. It's consistently turning sources into atomic notes.

Niklas Luhmann wrote 90,000 notes by hand over 40 years. You have articles, papers, and blog posts. The gap between reading them and having them in your vault is where most PKM systems collapse.

What "atomic" actually means

An atomic note has:

One concept (not one article, one chapter, or one idea cluster — one concept)
A title that could be a Wikipedia article
[[wikilinks]] to related concepts
A source citation

Bad: "Notes from The Shallows by Carr"

Good: "Neuroplasticity", "Deep Reading Mode", "Hyperlink Distraction Effect" (three separate notes)

The problem with manual atomization

Reading a 4000-word article and extracting 5 atomic notes takes ~25 minutes if you're doing it properly:

Read fully
Identify concepts worth keeping
Title each note (harder than it sounds)
Write the body — in your own words, not copy-paste
Add frontmatter, tags, wikilinks
Add source

Most people skip this entirely or write one large "literature note" that never gets revisited.

How Qonspekt automates this

Qonspekt takes an article and produces atomic notes using Claude AI:

Paste article → Claude extracts concepts → atomic .md files → ZIP download

The system prompt specifically instructs Claude to:

Extract 3–7 key concepts (not summaries — named, reusable concepts)
Write each as a self-contained note
Connect concepts via [[wikilinks]] within the batch
Never copy-paste — always paraphrase and synthesize

The wikilink constraint is important. Luhmann's system worked because notes referenced each other. Qonspekt replicates this: the "Spaced Repetition" note from one article will wikilink to the "Active Recall" note from the same article, creating a small interconnected cluster you can drop into your vault.

What to do after importing

Open the ZIP, check the notes
Edit anything that's wrong or imprecise
Add wikilinks to existing notes in your vault ([[existing-concept]])
Move to the right folder

The AI does the scaffolding. You do the integration.

Is this "cheating"?

Depends on your goal. If you're using a Zettelkasten to develop original ideas, the thinking happens when you read and select which articles to atomize, edit the generated notes, and integrate them with your existing vault.

Qonspekt handles the mechanical part. The intellectual work is still yours.

Try it

→ qonspekt.github.io/qonspekt

BYOK (bring your own Claude API key). No account, no backend. ~$0.003 per article.

The 30-second Obsidian note workflow (using Claude AI)

Qonspekt — Tue, 12 May 2026 17:50:15 +0000

If you use Obsidian, you've probably felt this: you read a great article, tell yourself you'll take notes later, and then... never do.

The problem isn't motivation. It's friction. Opening a new note, figuring out what to title it, structuring the frontmatter, extracting the key ideas, writing atomic notes one by one — it takes 20 minutes minimum for a good article.

I built Qonspekt to remove that friction completely.

The workflow

Read an article. Find one worth keeping.
Copy the text. Paste into Qonspekt.
Hit "Generate". Wait ~5 seconds.
Download ZIP. Drag into Obsidian.

Done. You now have 3–7 atomic notes, each covering one concept, each with proper frontmatter and [[wikilinks]] connecting related ideas.

What makes a good Qonspekt note?

Here's an example output from an article about the Feynman Technique:

---
title: "Feynman Technique"
tags: [learning, mental-models, teaching]
aliases: ["Feynman Learning Method"]
---
# Feynman Technique

A learning method developed by physicist Richard Feynman:
explain a concept as if teaching it to a child. When you
can't explain it simply, you've found a gap in your
understanding.

Related: [[Active Recall]], [[Spaced Repetition]]

## Sources
- https://example.com/feynman-article

Notice: one concept, clear title, 2–4 tags, wikilinks to related concepts from the same article, source citation.

Which articles work best?

High value:

Long-form essays (3000+ words) — lots of concepts to extract
Academic summaries — dense with named concepts
Technical tutorials — methods, patterns, terminology
Book summaries — ready-made concept clusters

Lower value:

News articles — facts, not concepts
Opinion pieces — hard to extract reusable knowledge
Short posts under 500 words — not enough to atomize

The sharing trick

After generating, click Share. Qonspekt encodes your notes into the URL. Send the link to a study partner — they see your notes instantly, no account needed.

This makes it useful for:

Sharing reading notes with colleagues
Building a shared PKM entry point for teams
Demonstrating your note-taking approach to others

Cost

~$0.003 per article with Claude Haiku 4.5. A $5 Anthropic credit covers ~1,600 articles.

New accounts get free credits. You'll probably never pay anything.

Try it

→ qonspekt.github.io/qonspekt

No account. No installation. Paste an article, get notes.

I built a BYOK browser tool that turns any article into atomic Obsidian notes

Qonspekt — Tue, 12 May 2026 17:23:09 +0000

I kept procrastinating on taking notes from articles I read. The cycle was always the same: read something interesting, tell myself I'll write proper notes later, never do it. So I built a tool to do it automatically.

What Qonspekt does

Paste any article → Claude AI extracts the key concepts → you get 3–7 atomic Markdown notes, ready to drag into Obsidian.

Try it: https://qonspekt.github.io/qonspekt/

Each note gets:

YAML frontmatter (title, tags, aliases)
[[wikilinks]] connecting related concepts from the same article
## Sources section with the original URL
Proper filename (concept-slug.md)

The technical approach: single HTML file, no backend

The whole thing is one HTML file. No framework, no build step, no server.

const response = await fetch('https://api.anthropic.com/v1/messages', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'x-api-key': userApiKey,
    'anthropic-version': '2023-06-01',
    'anthropic-dangerous-direct-browser-access': 'true',  // official BYOK header
  },
  body: JSON.stringify({ model, max_tokens: 4096, system, messages })
});

The anthropic-dangerous-direct-browser-access: true header is Anthropic's official way to support BYOK browser tools. Your key goes from your browser directly to Anthropic — I have no server to log anything.

Getting consistent JSON from Claude

The trickiest part: Claude sometimes wraps output in markdown code fences even when told not to.

const raw = data.content?.[0]?.text || '';
const match = raw.match(/\[[\s\S]*\]/);  // extract JSON array regardless of wrapping
if (!match) throw new Error('Could not parse response');
const notes = JSON.parse(match[0]);

This regex approach handles both clean JSON and fenced output reliably.

Serverless sharing

After generating, there's a "Share" button that base64-encodes all notes into the URL hash:

const enc = btoa(unescape(encodeURIComponent(JSON.stringify(notes))));
const url = location.origin + location.pathname + '#shared=' + enc;
navigator.clipboard.writeText(url);

On load, if the hash starts with #shared=, it decodes and renders the notes in read-only mode. No server, no database, no account needed for the recipient.

The system prompt

Getting the right structure took iteration. Key constraints that worked:

- Extract 3-7 key concepts, write one atomic note per concept
- Each note is self-contained but uses [[wikilinks]] to reference 
  related concepts from the same batch
- 150-280 words per note body
- Tags: 2-4, lowercase, hyphens for spaces
- Return ONLY a valid JSON array, no markdown fences

The "same batch" constraint for wikilinks is important — it makes the notes interconnected without hallucinating links to notes that don't exist.

Cost

~$0.003 per article with Claude Haiku 4.5. New Anthropic accounts get free credits.

Source

MIT licensed: https://github.com/Qonspekt/qonspekt

Would love feedback on the note structure or the prompt. What frontmatter fields do you typically use in your vault?