Chrome extension that condenses web pages using LLMs. Cut through the slop, keep the signal.
- Inline mode (default) — condenses each paragraph in place, keeping the original page layout. Links are preserved intelligently.
- Overlay mode — full-page summary with structured headlines and bullets
- Density slider — five levels from FLUFF (~20% reduction) to BARE BONES (~80% reduction), plus a Smart mode that lets the model decide per-paragraph
- Auto-condense — optionally runs on every page load, with a domain exclude list (supports wildcards like
*.google.com) - Parallel processing — long pages are chunked and condensed in parallel for speed
- Lifetime score — tracks cumulative words cut, mapped to book equivalents ("The Old Man and the Sea", "War and Peace", etc.)
- Keyboard shortcut — Option+T (Mac) / Alt+T (Windows/Linux)
- Clone this repo or download the ZIP
- Go to
chrome://extensionsand enable Developer Mode (top right) - Click Load unpacked and select the repo folder
- Click the TLDR icon in the toolbar — the first-run screen will walk you through setup
On first launch, pick your provider (OpenAI, Anthropic, or a custom OpenAI-compatible endpoint), paste your API key, and click I Agree · Start Cutting. Settings are saved locally and never leave your browser.
| Provider | Default model | Other options |
|---|---|---|
| OpenAI | gpt-4o-mini | gpt-4o, gpt-4.1-mini, gpt-4.1-nano |
| Anthropic | Claude Haiku 4.5 | Claude Sonnet 4.5, Claude Sonnet 4.6 |
| Custom | — | Any OpenAI-compatible endpoint (Ollama, Together, etc.) |
Use at your own risk.
- Page content is sent to a third-party API. Do not use on pages with sensitive information unless you understand and accept this.
- Auto-run sends every page you visit. A warning is shown when you enable this. The extension requests additional permissions only when auto-run is turned on. You can exclude specific domains.
- API keys are stored locally in
chrome.storage.local(on-device only, not synced to Google's cloud). - LLM output can be influenced by page content (prompt injection). The extension escapes all output before rendering, preventing code execution, but summaries could be misleading on adversarial pages.
- API costs are on you. Each condensation makes one or more API calls (long pages are chunked into parallel requests). There is no rate limiting.