Verdict first
+One action pulls the lede out of long articles, tickets, and threads so you can bail early or read with intent.
+Product
+One action pulls the lede out of long articles, tickets, and threads so you can bail early or read with intent.
+Follow-ups reference the page and the summary you already generated—no free-floating chat pretending it read for you.
+Ollama native, OpenAI-compatible hosts, Groq, LM Studio—pick the model and base URL in settings instead of a locked vendor.
+FAQ
++ Run ollama serve, confirm models with ollama list, and keep + Native mode on http://localhost:11434 without /v1. +
++ Wrong API mode for the URL: Native → no /v1. OpenAI-compatible → include + /v1 (for example http://localhost:11434/v1 for Ollama + compatibility mode). +
+Page summarizer and chat · Chrome and Firefox
++ Summarize and ask about this tab before you spend twenty minutes on noise. +
++ Answers trace the page you have open—not a blank assistant waiting for a prompt. +
+ +Keyboard
++ Ctrl + Shift + U + + ⌘ + Shift + U on macOS — + Summarize with Lede +
+Install
+
+ Clone webai-summarizer on Tangled—not this marketing repo. You need a folder that contains
+ manifest.json. lede-website is only this page.
+
Download or clone the extension repo so the project lives on disk.
+ Open Tangled repo +Chrome uses unpacked load; Firefox uses a temporary add-on.
+Point settings at Ollama native or an OpenAI-compatible endpoint, then try a tab.
+manifest.json+ Temporary add-ons clear when you fully quit Firefox—reload during active dev sessions. +
++ If you already use split manifests, ./build.sh chrome or + ./build.sh firefox from the extension repo targets a specific build. +
++ Source: + {tangled} +
+Privacy and APIs
+
+ /api/generate on your machine. Base URL omits /v1.
+
+ /v1/chat/completions to the endpoint you configure. Content leaves the browser only when
+ you summarize or chat—not via a script on every site.
+
Extraction is on-demand when you ask.
+In the popup
+Article, doc, issue—whatever you need to judge quickly.
+Toolbar icon, context menu, or the keyboard shortcut.
+Summarize for the lede, or type a question about this tab and press Enter.
+- Page summarizer & chat · Chrome & Firefox -
-- Put the main point first—then ask anything about the page—with AI. Answers stay tied to - the tab you have open, not a generic chat shell. -
- -- Keyboard -
-- Ctrl + Shift + U - - ⌘ + Shift + U on macOS - — Summarize with Lede -
-Install
-
- Clone the extension, not this site. Lede's source and manifest.json
- live in the webai-summarizer repo on Tangled. This repository (lede-website) is only the marketing page.
-
Start here
-
- Download or clone webai-summarizer from Tangled so you have a local folder
- with manifest.json.
-
Chrome (unpacked) or Firefox (temporary add-on). Detailed clicks are in the panels below.
-- Point settings at Ollama native or an OpenAI-compatible endpoint, then summarize any tab. -
-manifest.json)- Temporary add-ons last until you fully quit Firefox—plan to reload during active development. -
-- In the extension repo, you can target a browser with ./build.sh chrome or - ./build.sh firefox if you maintain separate manifests—only if you already - use that workflow. -
-- Open source: - {tangled} -
-- Worth your time? Lede reads the open tab on demand—then puts the verdict up front. --
Product
-In the popup
-Whatever you need to understand—article, ticket, README.
-Toolbar icon or shortcut.
-Summarize for the lede, or type a question about this tab and press Enter.
-Privacy & APIs
-
- Uses /api/generate. When the model runs locally, content stays on your
- machine. Base URL should not include /v1.
-
- Uses /v1/chat/completions for cloud or self-hosted endpoints you configure.
- Page text is sent only when you summarize or chat—not from a script that runs on every site.
-
- Extraction is on-demand when you ask—nothing leaves the tab until you trigger it. -
-FAQ
-- Run ollama serve, check ollama list, and keep Native - mode on http://localhost:11434 without /v1. -
-- Wrong API mode for the URL: Native → no /v1. OpenAI-compatible → include - /v1 (for example http://localhost:11434/v1 for Ollama - compatibility mode). -
-