AnswerIT is a userscript that combines tab-switch detection bypass techniques with a configurable multi-provider AI helper. It lets you navigate away from assessment tabs without triggering warnings and offers a modern, intuitive interface to generate answers using any AI provider of your choice.
- Neutralizes browser visibility APIs and blocks events (e.g., "blur", "visibilitychange") so you can safely switch tabs during assessments.
- Supports multiple OpenAI-compatible providers.
- Streaming responses (see Tampermonkey note below).
- Per-model settings (
reasoning,image input, extra JSON options). - Caching + regeneration history.
- Optional auto-run on question change.
- Popup UI with drag, resize, snap, theme toggle, output mode toggle.
- Dedicated config page for providers, models, preferences, and supported sites.
- Provider and model chip bars for quick add flows.
- API key test button per provider.
The script includes platform selectors for multiple learning/assessment websites. The configure page also shows detected supported sites.
-
Install a userscript manager
- Violentmonkey (recommended)
- Tampermonkey
-
Install the script
-
Open configure page (first time setup is mandatory)
Note:
Tampermonkey Bug: Due to a known bug in Tampermonkey on Chromium browsers, model responses will not stream in real-time (you'll only see the final answer after generation). For full streaming support, use Violentmonkey.- As of v5+, this issues is presumably fixed, but its all only a workaround (figuring it out was a pain) and until a proper fix is out, Violentmonkey is still the recommended userscript manager
Built-in presets currently include:
- Gemini (free)
- Groq (free)
- OpenRouter (free)
- Nvidia NIM (free)
- OpenAI (paid)
- Anthropic (paid)
- Azure OpenAI (paid)
- Ollama (free?)
- Z.AI (free)
- Kilo (free)
You can also add custom providers.
(btw, free doesn't mean 100% free... obviously there are limits 😮💨)
- Add models per provider.
- Reorder by drag handle.
- Configure:
- display name
- description
- reasoning toggle
- image-input toggle
- advanced request JSON options
- Hotkey: default
Alt + A - Popup theme:
lightordark - Default output view:
rawormarkdown - Models to show: first
4or8(last visible slot is flexible) - Editable output (raw mode)
- Detection Bypass: Neutralizes tab-switch detection on supported platforms
- Question Detection: Finds and processes questions (text, images, code)
- Model Selection: Choose Gemini, OpenAI, or Anthropic
- Streaming Response: (If supported) See answers as they generate
- Direct Insertion: Insert answers with one click
- API Key Issues: Double-check for typos, extra spaces, or expired keys
- Provider test fails: verify endpoint + key + model access.
- No model buttons shown: ensure at least one provider is enabled and at least one model is enabled.
- Streaming seems delayed: prefer Violentmonkey.
- UI missing: refresh page and verify script is enabled.
- Report issues: GitHub Issues
- Keys/settings are stored locally via userscript storage.
- The project does not run a hosted backend for your prompts/answers.
- You can inspect source code directly on GitHub.
This project is released under the MIT License.
This project is provided for educational and research purposes only. It serves as a proof-of-concept to illustrate browser automation techniques and AI integration patterns.
Important Notes:
- Understanding these techniques helps developers build more robust security measures
- Use responsibly and ethically in accordance with your institution's policies
- Respect the terms of service of the platforms you use
- The developer disclaims liability for any misuse or consequences arising from use
Academic Integrity: This tool should be used to supplement learning, not replace it. Always prioritize understanding concepts over simply obtaining answers.
Tip: For the best experience, use Violentmonkey on Chromium browsers.