A cross-platform tool that captures your current browsing context (page URL, title, selected text) plus a clipboard fallback, formats it into a clean markdown block, and copies it to your clipboard whis is ready to paste into an LLM chat. Includes a small local history so a capture isn't lost if you copy something else before pasting.
Development on Context Capsule v1.0 is finished.
The final v1.0 build delivers that core flow: select text on a Chromium page, open the extension popup, click Capture Current Page, and paste a clean markdown block with the page title, URL, timestamp, and selected text.
If no text is selected, Context Capsule uses the current clipboard as a fallback so the capture still works.The browser extension stays intentionally thin. It collects page context and talks to the local Python host through Chrome Native Messaging. Python owns the durable logic: formatting, clipboard writes, SQLite history, capsules, exports, diagnostics, and install registration.
install.py for Windows, macOS, and Linux.docs/install.md.For now, the core product is complete: Context Capsule turns a multi-step context-sharing routine into one popup click, keeps a recoverable local history, and produces clean markdown ready for an AI chat.
Built the following in my tool Context-Capsule:
Use the popup’s Capsule section to collect multiple captures before pasting into an AI chat:
Start: creates a new active capsule.Append Page: captures the current page and adds it to the active capsule.Copy: copies all capsule items as one combined markdown prompt.Clear: removes the active capsule.The extension options page stores browser-side defaults in chrome.storage.local:
Use the popup Project and Tag fields before capturing. Labels are stored with history entries and capsule items, shown in entry metadata, and searchable through the project/tag filter.
The popup Export controls copy generated export text to the clipboard:
Visible: exports the currently visible filtered history rows.All: exports all saved history rows.Capsule: exports the active multi-source capsule.The host stores a capped local diagnostics log for recent captures, exports, capsule actions, and errors. The popup Diagnostics section shows the latest entries so native messaging and clipboard failures are easier to debug.
The popup shows a first-run Quick Start card until dismissed. It links to the localhost demo page and reminds users about the capture/history hotkeys and doctor command.
The popup Demo Prompt button copies a polished prompt using the active capsule when one exists, otherwise the latest three captures. This is designed for quick hackathon demos and judge walkthroughs.
Serve demo.html through Python’s built-in local server for a controlled manual test page with selectable sample text and checklist prompts: python -m http.server 8765
Hey Guys, I just started a new project Context Capsule.
Context Capsule cuts the “share what I’m looking at with an AI” routine from a 5-6 step manual dance down to a single hotkey.
When you’re debugging, researching, or working with an AI assistant, you constantly need to hand it context: which page you were on, what part of it matters, plus whatever you’d already copied.
Right now that means selecting text, copying it, switching to your chat window, pasting it, going back to grab the URL, and then cleaning everything into a readable shape.
If you accidentally copy something else before you paste, the original context is gone and you have to redo the whole thing.
Context Capsule captures the active Chromium page URL, title, selected text, and a clipboard fallback, formats it as markdown, copies it to your clipboard, and keeps a small local SQLite history.
