Why?
Most “AI from any page” tools proxy your data or make you buy credits. I wanted something transparent that:
- stays in your browser,
- uses the official provider UI,
- and doesn’t require another wallet of tokens.
How it works
1. Right-click -> PageToChat.
2. A sidebar appears where you choose:
- Provider & model
- Chat mode: temporary (ephemeral) or persistent
- Prompt (or pick from saved ones)
- Attachments (auto-detected):
* page screenshot
* page content (with/without HTML)
* selected text (if you had text highlighted)
* image (if you clicked on an image)
* site-specific extras (today: Twitter/X threads and Gmail email threads)
3. Hit Send → we open the provider tab, pick the model, enable temporary mode if needed, fill the prompt, upload attachments, and (optionally) auto-submit.
"Quick actions" (custom context menu items)
You can save presets—provider, model, prompt, and which attachments to include—and they show up as extra right-click items. Clicking one fires the whole flow automatically.
Pricing
- Free: all features, 1 saved quick action, occasional small sidebar ad (no page context sent to ads).
- Power: multiple quick actions, no ads. (Feedback welcome on this split-happy to tweak based on what HN finds fair.)
What makes it different
- No middleman credits; uses your existing LLM account.
- Zero copy-pasting between tabs.
- Context-aware attachments (and growing list of site-specific integrations).
What I’d love feedback on
- Missing providers/models you want first.
- Additional site-specific attachments (Docs? GitHub issues/PRs? Notion pages?).
- Default attachment choices you’d expect for common workflows.
Happy to answer anything - privacy, permissions, roadmap, bugs. If you try it, I’d love to hear what breaks on your setup and what would make it a daily driver.