Glasp

Injects a browser extension overlay into web pages and YouTube video players that enables users to select and highlight text/sections with customizable colors. The extension uses DOM mutation observers to track page changes and maintains highlight state in the browser's local storage, syncing selections across page reloads. Highlights are stored with metadata including URL, timestamp, and color tag for later retrieval and organization.

Provides a personal dashboard interface that aggregates all highlights across sources into a searchable, filterable library. Uses a tag-based taxonomy system and color-coded categorization to organize highlights by topic, source, or custom metadata. The library supports full-text search across highlight content and source URLs, with sorting by date, source, or color tag. Highlights can be grouped into custom collections or folders for thematic organization.

Processes selected highlights or entire collections through an LLM API (likely OpenAI or similar) to generate concise summaries, key takeaways, or thematic synthesis. The extension batches highlights by source or collection and sends them to the backend with context about the original article/video, receiving structured summaries that are cached and displayed in the library. Summaries are regenerable and can be customized by summary type (bullet points, paragraph, key quotes).

Enables users to publish highlights and collections to a public or semi-public community feed where other Glasp users can discover, upvote, and follow curators. The backend maintains a social graph of follower relationships and uses engagement signals (upvotes, saves, shares) to rank highlights in discovery feeds. Users can browse highlights by topic, trending curators, or follow specific users to see their new highlights. Shared highlights include attribution to the original curator and link back to the source article/video.

Exports highlights from the Glasp library to external tools and formats including CSV, JSON, Markdown, and direct integrations with Notion, Obsidian, and other note-taking apps. The export pipeline preserves metadata (source URL, timestamp, color tag, collection) and formats highlights according to the target tool's expected structure. For native integrations (Notion, Obsidian), the extension uses their respective APIs to create new pages or notes with highlights automatically organized by collection or source.

Detects the type of content being highlighted (article, YouTube video, academic paper, blog post) and extracts relevant metadata including title, author, publication date, video duration, and thumbnail images. For YouTube videos, the extension captures the video ID and timestamp of highlighted sections, enabling users to jump directly to relevant moments. For articles, it extracts the article text, byline, and publication metadata. This metadata is stored alongside highlights to provide rich context in the library.

Stores not just the highlighted text but also surrounding context (previous and following sentences/paragraphs) from the original source, enabling users to understand the highlight's meaning without revisiting the source. When viewing a highlight in the library, users can expand to see the full context window. The extension uses DOM traversal to capture paragraph-level context at highlight time and stores it alongside the highlight text. Context is searchable and can be included in exports.