History

3 snapshots

Tracking since May 12, 2026.

View as table

Date	Users	Rating	Reviews	Version
May 12, 2026	—	—	—	0.1.0
May 17, 2026	—	—	—	0.1.0
May 31, 2026	1	—	—	0.1.0
Now	2	—	—	0.1.1

Changelog

May 31, 2026

description

OutlineMySource builds a structured Markdown outline from content you already have—so you can scan the shape of an article, PDF, pasted notes, subtitles, or a local audio/video file before investing deep reading or editing time.

How it works

Active tab (websites): Uses a Readability-style extraction path when it finds a clear article, then falls back to visible plain text. Everything runs in your browser on your machine.
Local files: Pick PDFs, UTF‑8 text, subtitles (.srt / .vtt), or audio/video from disk. Your files are not uploaded to our servers; decoding and speech recognition run locally in Chrome (Transformers.js / ONNX where applicable).
Outline format: You get Markdown with an optional compact metadata block—easy to save, diff, or drop into notes and AI tools.
Why install it

Skim-first workflow: See headings and spaced excerpts instead of drowning in full pages or long transcripts.
Offline-minded: No account required; processing stays local. Network may be used only when first loading optional ML assets (e.g. speech models from public CDNs) into the browser cache—similar to other local ML extensions.
Export: Download the outline as .md with one click.
Languages

Panel available in English and Spanish (selector in the extension UI).
Good to know

Outlines are extractive (snippets from the source), not a replacement for careful reading or professional transcription services.
Long captures may be clipped to a generous character limit shown in the export.

OutlineMySource builds a structured Markdown outline from content you already have—so you can scan the shape of an article, PDF, pasted notes, subtitles, or a local audio/video file before investing deep reading or editing time.

How it works

• Active tab (websites): Readability-style extraction when the page has a clear article, then a visible-text fallback. Processing runs in your browser on your device.

• Chrome built-in AI (when available): On supported Chrome versions, the extension can use Chrome’s on-device Summarizer API (Gemini Nano) to produce a concise synthesized outline from web pages and pasted text. If the API is unavailable or not ready, it automatically falls back to an extractive outline (quoted snippets from the source).

• Local files: Open PDFs, UTF-8 text, subtitles (.srt / .vtt), or audio/video from disk. Files are not uploaded to our servers. PDF parsing, speech recognition (Whisper via Transformers.js / ONNX), and optional semantic indexing run locally in Chrome.

• Audio & video: Local transcription with progress and cancel. Long jobs use a background worker so the panel stays responsive. When Chrome’s Summarizer is available and transcription quality is good, the main outline may use on-device synthesis on the transcript; otherwise you get an extractive outline with clear metadata in the export.

• Outline format: Markdown with a compact YAML metadata block (platform, engine used, fallbacks)—easy to save, diff, or paste into notes and AI tools.

Why install it

• Skim-first workflow: See headings and spaced excerpts (or a short synthesis when Chrome AI is on) instead of drowning in full pages or long transcripts.

• Privacy-minded: No account required. Core processing stays on your machine. Network may be used only when first loading optional ML assets (e.g. speech models from public CDNs) into the browser cache—similar to other local-ML extensions. Summarizer mode uses Chrome’s built-in on-device model, not AI4Context servers.

• Export: Download the outline as .md with one click.

Languages

Panel UI in English and Spanish (language selector in the extension).

Good to know

• Outlines are meant to help you skim and orient—not to replace careful reading, legal review, or professional transcription.

• Chrome’s Summarizer requires a compatible Chrome channel and user gesture; availability varies by device and Chrome version.

• Long captures may be clipped to a generous character limit noted in the export metadata.

• The Microsoft Edge listing is a separate build without Chrome’s Summarizer; it uses extractive outlines plus the same local Whisper/MiniLM media path.