LLM-powered Chrome extension that filters low-value, distracting, and manipulative content from the web. Tell it what you don't want to see and dont see it. so many people are fighting for your attention its time for you to fight back.
you can get the chrome extention here https://chromewebstore.google.com/detail/unwired/eagjafndbcedibfalnfimildfphokffn
- Open Chrome and go to
chrome://extensions/ - Enable Developer mode (top right)
- Click Load unpacked and select the
extension/folder - Click the Unwired icon in the toolbar
- Add your API key in Settings (supports OpenAI, Claude, Gemini, Grok, OpenRouter)
- Write your rules and start browsing
No backend server needed. The extension calls LLM APIs directly from the browser using your own API key.
No clickbait
No politics
Only programming content
No shorts or playables
No ads or sponsored content
No outrage bait or drama
Highlight educational content
Dim entertainment
Chrome Extension (Manifest V3)
┌──────────────────────────────────┐
│ content.js │
│ Extracts content from the DOM │
│ Batches items for classification│
│ Applies hide/show/dim/highlight │
│ │
│ background.js │
│ Calls LLM API directly │
│ Classifies content per rules │
│ Refines rules when you reject │
│ │
│ popup.html/js │
│ Rules editor (auto-saves) │
│ Settings & API key │
│ Recently removed & history │
└──────────────────────────────────┘
| Site | What gets scanned |
|---|---|
| YouTube | Video titles, Shorts shelves, Playables, ads, comments |
| Twitter/X | Tweets, promoted tweets, app install ads, Who to Follow |
| Search results, People Also Ask, sponsored results | |
| Post titles | |
| Feed posts | |
| Feed posts | |
| Any site | Generic extractor (articles, cards, headlines, table rows) |
| Action | Behavior |
|---|---|
show |
Left unchanged |
hide |
Removed from DOM entirely |
dim |
Faded out, visible on hover |
highlight |
Blue border, stands out |
- Natural language rules — write filters in plain English
- Auto-save — rules apply as you type (800ms debounce)
- Smart reject button — hover any content to see an X button; clicking it sends the content to the LLM which analyzes the category and rewrites your rules
- Prompt history — every rule change is saved, restore any previous version
- Filter button toggle — enable/disable the hover reject buttons in Settings
- Multi-provider — OpenAI, Claude, Gemini, Grok, or OpenRouter
- No API key mode — basic keyword matching fallback
- Per-site control — enable/disable filtering per site, or turn on for all sites
- Google app protection — Colab, Docs, Drive, etc. are never filtered
| Provider | Default Model | API Key Format |
|---|---|---|
| OpenAI | gpt-4o-mini | sk-... |
| Anthropic | claude-haiku-4-5 | sk-ant-... |
| Gemini | gemini-2.0-flash | AIza... |
| Grok | grok-3-mini-fast | xai-... |
| OpenRouter | openai/gpt-4o-mini | sk-or-... |
- Batches up to 25 items per API call
- In-memory cache (500 entries) prevents re-classification
- Debounced scanning (400ms) avoids excessive API calls
- MutationObserver handles infinite scroll and SPA navigation
- Stale results are discarded when rules change mid-flight