Sparx AI Tools is a Chrome extension that adds homework help tools to Sparx Reader and Sparx Science pages.
Warning
Due to the nature of the sparx science extension controlling your browser to automate question answering, Microsoft Defender may flag the files as a false positive. To solve this, allow it through defender.
Reader mode adds quick clipboard helpers directly into the page UI:
- Simple design: Just hit copy and paste into whatever LLM you prefer
- Book copy button: copies the current reading text into a prepared prompt format
- Question copy button: copies the active question + options into a prepared prompt format
LLM: Gemini by google (?)
Science mode provides a one-click workflow around question extraction and AI response handling:
- Solve Science button for one-click processing
- Stable question extraction from Sparx question parts and tables
- Gemini tab orchestration (find/open, pin, submit, wait for response)
- Structured response parsing with guarded start/end markers
- Floating result panel with formatted answers (
a),b), etc.) and copy/close actions - Global menu entry:
Sparx AI Toolsin the Sparx header menu - Settings modal for timeouts/retries and capture behavior
- Auto-recovery on transient timeouts (refresh/retry path)
From current testing:
If you test another model and want it listed here, open an issue.
- Download the latest release ZIP from the Releases page.
- Extract the ZIP to a normal folder.
- Open
chrome://extensions. - Enable Developer mode.
- Click Load unpacked and select the extracted folder (not the ZIP file).
- Make sure the extension is enabled.
- Open a Reader page.
- Use the copy buttons on book/question panels.
- Paste into your preferred tool.
- Open a Science question page.
- Click Solve Science.
- Use the floating response panel to review/copy the output.
- Open Menu -> Sparx AI Tools to tune retries/timeouts if needed.
The extension may work with other AI models, but I have only tested it with gemini
I used gemini due to the high limits on images and calculations.
If you test another model and it works well, open an issue and I'll add it as an official model.




