Tangentry

Project Overview

Tangentry is a local-first tool that aggregates content from Bilibili, YouTube, Xiaohongshu, Douyin, Xiaoyuzhou podcasts, RSS, and web articles into one interface. Videos and podcasts become readable: subtitle highlighting, full-text search, bilingual display, and AI summaries in one place. Notes anchor to timestamps, quoted passages, PDF locations, or screenshots and export as Markdown ready for Obsidian. All data lives in a local SQLite database with no account required.

Item	Value
Platform	macOS, Windows, Linux, Docker
Stack	Go 1.27, React 19, SQLite
Port	7380 (server mode)
License	MIT

Read subtitles while watching, take notes with screenshots — all data stays on your own computer.

Key Features

• Aggregated discovery: recommendations from Bilibili, YouTube, Xiaohongshu, Douyin, and Xiaoyuzhou in one feed with adjustable source preferences and a read-later queue
• Subscriptions: follow uploaders, channels, RSS, podcasts, and linux.do topics with groups, inbox, scheduled refresh, and OPML import/export
• Link parsing: paste a link or ID to add videos, uploaders, podcast episodes, web articles, or local PDFs
• Subtitle reading: video on the left, subtitles on the right with highlighting, click-to-jump, and in-subtitle search
• Transcription: local whisper.cpp and faster-whisper or cloud services like Groq, with platform subtitles preferred
• AI assistant: connect any OpenAI-compatible API for video and article summaries plus bilingual subtitle and text translation
• Notes: timestamp, highlight, screenshot, and PDF anchor notes with nested tags, full-text search, and Obsidian-ready Markdown export
• Downloads and favorites: yt-dlp based download queue and Bilibili favorites sync with resume support
• Multi-platform: native desktop apps or Docker self-hosting on a NAS, accessed from the browser

Target Audience

• Knowledge workers
• Podcast listeners
• Note-taking and Obsidian users
• NAS self-hosting enthusiasts

Quick Start

1. Download the installer for your platform from Releases; desktop builds bundle the runtime, and yt-dlp plus transcription models download automatically on first use
2. Sign in to Bilibili via QR code and configure AI or transcription services in Settings
3. Paste any supported link on the Add page, or add RSS feeds and uploaders under Subscriptions
4. Take notes while reading — timestamps are attached automatically — and export Markdown ZIP for Obsidian

Core Advantages

• Local-first: all content, notes, and history live in local SQLite with no account, no telemetry, and credentials never leave the device
• Reading-first: videos and podcasts become readable with subtitle tracking, full-text search, and bilingual display
• Notes as assets: notes anchor to timestamps or passages, jump back to the source, and export in open formats

System Requirements

Item	Requirement
Desktop	macOS, Windows, Linux (bundled runtime)
Docker	Any Docker host, port 7380
Source build	Go 1.27+, Node.js 24+
Optional	yt-dlp, ffmpeg (auto-downloaded on first use)

Notes

• Early stage: the project is in 0.x; interfaces and data structures may change, with automatic database migration
• Platform changes: video parsing and downloading rely on yt-dlp; update the tools in Settings when platforms change
• Public deployment: server mode uses a single shared password; put an HTTPS reverse proxy in front before exposing it publicly
• Content copyright: fetched content belongs to its platforms and authors; do not mass-scrape, redistribute, or use commercially