capturemd

A personal and very idiosyncratic productivity tool for capturing web content into structured markdown notes with support for video downloads through yt-dlp. Connects to self hosted services like Wallabag and FreshRSS. Meant to be a replacement to org-capture living outside of Obsidan.

Features

• Smart URL Capture: Convert web pages, YouTube videos, GitHub repos, Reddit threads, Hacker News posts, Steam pages, Google searches, and podcasts into markdown
• Obsidian & Dataview Compatible: Notes include frontmatter with metadata for querying and organization
• Jellyfin Integration: YouTube videos cached as media files with NFO metadata (year as season, video as episode)
• Episode Reindexing: Automatically renumbers episodes chronologically after caching for Kodi compatibility
• Wallabag Integration: Share content from mobile to Wallabag, sync as markdown notes, visible in Dataview queries
• Google Search Workflows: Share search links to Wallabag, process on desktop into markdown inbox
• Browser Integration: Map Tridactyl or qutebrowser to capture bookmarks directly
• FreshRSS Support: Convert feed items to markdown notes
• Local Caching: Cache videos and content for offline access

Installation

pip install . # --break-system-packages 

Usage

From CLI

# Capture from clipboard
capturemd url

# Capture specific URL
capturemd url "https://example.com"

From browser

Similar as from the CLi, this can be called from Qutebrowser/Tridactyl withsomething like:

Tridactyl

Add to ~/.tridactylrc:

# Capture current page
bind ,b composite get_current_url | js -p tri.excmds.shellescape(JS_ARG).then(url => tri.excmds.exclaim_quiet('notify-send Captured && ~/.local/bin/capturemd url ' + url))
# Capture "hinted" link 
bind ,B hint -qW js -p tri.excmds.shellescape(JS_ARG).then(url => tri.excmds.exclaim('notify-send "Captured target link" && ~/.local/bin/capturemd url ' + url))

Qutebrowser

Add to ~/.config/qutebrowser/config.py:

# Capture current page
c.bind(',b', 'spawn capturemd url {url}')
# Capture "hinted" link
c.bind(',B', 'spawn capturemd url {hint-url}')

From Android

• wallabag is the main way I get data out of my Android phone (although it should work on iOS)
• sharing a link to wallabag will eventually have capturemd process it on a cron
• one off notes are entered as google search in a new tab and added into the "$SHARE/notes/browser_notes.md" file

Parsing

Parse from external services into markdown

capturemd parse-rss # parses the starred items
capturemd parse-wallabag # parses new entries from wallabag and tag them as parsed if meant to stay there, otherwise removes the entry

when parsed from external sources or manually from a URL, notes are initially bare and only specifiy a url/platform-specific UUID (like the ID of a youtube video). Then when runnin parse the script contacts the appropriate server and writes down the information.

# Parse all notes
capturemd parse

Environment Variables

capturemd can be configured using environment variables. Copy .env.example to .env and modify as needed.

Service Configuration

Variable	Description
WALLABAG_HOST	Wallabag instance URL
WALLABAG_CLIENT_ID	Wallabag OAuth client ID
WALLABAG_CLIENT_SECRET	Wallabag OAuth client secret
WALLABAG_USERNAME	Wallabag username
WALLABAG_PASSWORD	Wallabag password
FRESHRSS_URL	FreshRSS API endpoint
FRESHRSS_USERNAME	FreshRSS username
FRESHRSS_PASSWORD	FreshRSS password

Directory Configuration

By default, capturemd uses ~/share for notes and ~/Media for cached media. These can be overridden:

Variable	Description	Default
CAPTUREMD_SHARE_BASE	Base directory for notes and markdown files	~/share
CAPTUREMD_MEDIA_BASE	Base directory for cached media (videos, podcasts)	~/Media

Example:

export CAPTUREMD_SHARE_BASE=/data/share
export CAPTUREMD_MEDIA_BASE=/data/Media

Single file argument passing

Alternatively both the parse and the cache youtube sub-commands can take the path to a markdown file and target it instead of going over the entier share/vault.

You can do something like:

nvim $(capturemd url <some-youtube-url>)

(note that the "url" command only returns the full path to a markdown document whether it existed already or not)

And then from nvim:

:! capturemd parse %:p && capturemd cache youtube %:p &

to see the information populate and the video download in the background.