Coming Soon — TestFlight

Obsidian
Clipper

Clip any web page, text, or image into your Obsidian vault as clean Markdown — straight from the iOS Share Sheet. Image extraction and on-device OCR included. No plugin required.

Join the TestFlight View on GitHub

MIT licensed · Open source

Obsidian Clipper main settings screen showing vault, clipping options, and how-to-use steps

How it works

Three taps. Zero plugins.

Obsidian Clipper installs a Share Extension. Anywhere iOS lets you tap Share — Safari, X, Reddit, Mail, Messages, RSS readers — Clipper shows up alongside the system actions.

Ars Technica article 'Manufacturing qubits that can move' open in Safari on iPhone

Step 01

Open the page

Read the article like you normally would. In Safari, the extension uses live DOM preprocessing — what you see is what gets clipped, even behind logins or paywalls you're authenticated into.

iOS Share Sheet showing ObsidianClipper as a share destination alongside News, Reminders, and More

Step 02

Tap Share → Clip to Obsidian

The extension activates on URLs, plain text, HTML, and images. URLs embedded in shared text (X, Reddit, Messages) are auto-detected and fetched.

Clipped article rendered in Obsidian with YAML properties and Markdown body

Step 03

Find it in your vault

A self-contained folder lands in your chosen target folder: the Markdown note, an images/ subfolder, and YAML frontmatter ready to query in Dataview.

Features

Built like a real app.
Not a webview wrapper.

A native iOS Share Extension. Pure Swift. SwiftSoup for parsing, Apple Vision for OCR, Mozilla Readability–inspired scoring for extraction. Everything runs on-device.

📲

Share from anywhere

Activates on URLs, text, HTML, and images. Works in every iOS app with a Share button, not just Safari.

🧠

Live DOM in Safari

A JavaScript preprocessing file hands the extension the rendered page — including authenticated and JS-hydrated content. No re-fetch surprises.

⚡

JSON-LD fast path

When the publisher embeds a Schema.org Article block (Wired, NYT-style sites), the extractor uses the canonical body directly. ~30–40× faster than scoring.

📰

Readability fallback

When JSON-LD doesn't fire, a SwiftSoup-parsed scoring pass picks the article subtree and excludes navigation, ads, sidebars, cookie banners, and related-articles.

🖼️

Inline image extraction

Pulls srcset, data-src, data-lazy-src, data-original, and <picture><source>. Streams to disk so the extension stays under iOS's memory budget.

🔎

On-device OCR

Apple Vision recognizes text in clipped images. A coverage filter drops OCR output when text is incidental — no signage or license-plate noise.

📁

Self-contained clips

Each clip becomes its own folder with the Markdown note and an images/ subfolder. Move it in Obsidian and the images move with it.

🪶

Clean Markdown

Headings, lists, code blocks, blockquotes, tables, links, inline images — all rendered by a tree-walking Markdown converter, not a regex hack.

🔒

Privacy-first

No server, no analytics, no telemetry. Everything runs on-device. HTTPS-only ingress; javascript:, file:, ftp:, and data: URLs are rejected.

Output

Markdown that opens cleanly in Obsidian.

Each clip is a folder you can move, archive, or link from anywhere in your vault. YAML frontmatter is sanitized (no key injection from titles with newlines), filenames are sanitized (no Windows-reserved names slipping through), and image paths are relative.

Folder layout

<Vault>/
└── Inbox/
    └── Manufacturing qubits that can move/
        ├── Manufacturing qubits that can move.md
        └── images/
            ├── 3f2a-1.png
            └── 3f2a-2.jpg

Note contents

---
title: "Manufacturing qubits that can move"
source: "https://arstechnica.com/..."
clipped: 2026-05-08
type: article
---

# Manufacturing qubits that can move

> [Source](https://arstechnica.com/...) — Clipped 2026-05-08

To get quantum computing to work, we will
ultimately need lots of high-quality qubits…

![3f2a-1](images/3f2a-1.png)

## Extracted Text (OCR)

### 3f2a-1.png

> Recognized text from the image
> via Apple Vision…

Clipped article opened in Obsidian showing YAML properties and the rendered Markdown body

Obsidian Clipper settings in dark mode showing vault location, YAML frontmatter, save images, and OCR toggles

Privacy & safety

Your notes never leave your device.

No server. No analytics SDK. No telemetry. The extension fetches the page you shared, processes it on-device, and writes Markdown to the vault folder you chose. That's it.

On-device only

Extraction, OCR, and Markdown rendering all run inside the share extension. Nothing is sent to MMTAC or anyone else.
HTTPS-only ingress

file://, javascript:, ftp:, and data: URLs are rejected at every network boundary.
Memory-aware

Streams images to disk; drops concurrency on iOS memory warnings; downscales large images before OCR. Stays under iOS's ~120 MB extension budget.
Per-clip caps

20 images max, 50 MB cumulative cap, 15-second per-image timeout. 2 MB HTML cap.
Sanitized output

YAML values escape newlines, quotes, backslashes, tabs. Filenames strip invalid characters and Windows-reserved names.

Requirements

What you'll need.

iOS

17.0 or later

Vault

Obsidian on iCloud Drive or local Files

Cost

Free · MIT-licensed

Be one of the first to clip.

Obsidian Clipper is on TestFlight. Tap to install the latest beta — you'll need the TestFlight app on your iPhone or iPad.