8.7

Changelog (versions 8.0 to 8.7)

---

Version 8.0 (2025-04-12)
Comprehensive upgrade to the 8.x series—many new features and improvements:

• New feature: Added a “Display Answer Only” mode to distinguish it from the original auto-fill mode. Users can now switch between “Auto Fill” and “Display Answer Only”. In the latter, the script shows only the solution steps and final answer without injecting code to fill in answers. A “Rollback” button has also been added to restore the problem area to its pre-answer state (undoing any auto-filled changes).
  github.com github.com

• Improvements:

  • Enhanced logging and debug information. A log panel now appears in the UI, recording key operations and errors for easier tracing.
    github.com github.com
  • Token-usage statistics displayed at the top of the interface, showing the total API tokens used.
    github.com github.com
  • Refactored configuration data structure: unified storage in localStorage (with backward-compatibility migration), and support for saving different API keys and endpoints per model for easier multi-model management.

• Fixes: Resolved various known bugs to improve stability—for example, corrected an issue where log output was truncated in earlier versions.

---

Versions 8.1 & 8.2
These two version numbers do not appear separately in the public release history—likely internal builds or a naming hiccup. The developer quickly realigned the numbering to the main 8.x line. No standalone feature releases.

---

Version 8.3 (2025-04-14)
Building on 8.0’s foundation:

• New feature: Support for the new “chatgpt-4o-least” model—a GPT-4o variant fine-tuned with RLHF for more conversational output (though potentially less stable). Added to the model dropdown with explanatory notes.
  github.com

• Fixes: Corrected the version-tagging error (briefly mislabeled as v13.0, then reverted to v8.3). Polished the dual-mode switch to ensure reliable operation.

• Other: UI tweaks—updated certain labels and buttons for full bilingual (CN/EN) support, and added all new controls to the langText dictionary for internationalization. Marks the 8.x series entering its stable iteration phase.

---

Version 8.4
(skipped—no public commits; development proceeded directly to 8.5)

---

Version 8.5 (2025-04-18)
Feature expansion and important fixes:

• New feature: Enhanced model management—built-in “refresh model list” function reads new models from the OpenAI API’s /models endpoint and auto-adds them (e.g. gpt-4.1). A “Custom” option lets users manually enter a model name.
  github.com github.com
  This makes the script future-proof for new OpenAI releases.

• Fix: LaTeX output formatting—fixed issues in math formula parsing and rendering. Previously, expressions like $-\frac{3}{10}$ might display incorrectly (e.g. with stray dashes). Added wrapLatex and unescapeLatexDollar functions to preprocess GPT output into correctly wrapped Markdown LaTeX, and enforced that <answer>…</answer> only contain plain text or LaTeX (no Markdown symbols) to eliminate rendering errors.
  github.com github.com

• Improvement: Progress indicator refined—the simulated progress bar during requests now reliably resets, giving users clear feedback on script status.

---

Version 8.6 (2025-04-19)
Minor update, mainly consolidating previous changes:

• Fix: Continued refinement of math-formula rendering from v8.5, ensuring all problem types (with or without images/LaTeX) parse and display correctly.

• Improvement: Model-configuration stability—addressed edge cases where switching to a new or custom model could fail to update the endpoint or API-key fields. Now, model switches immediately sync both UI and config storage.

---

Version 8.7 (2025-05-01/07)
An integration-and-polish release for the 8.x series, focusing on UI experience:

• New feature: Collapsible settings panel—added a toggle on the “Settings” button to expand/collapse advanced settings (API Key, model selection, etc.), defaulting to collapsed to reduce UI clutter.
  github.com

• Improvements:

  • Layout optimization: when GPT returns long answers, the content area now auto-scrolls with a scrollbar to ensure full readability. Adjusted CSS to prevent overlap between the log and answer areas.
    github.com
  • Tweaked panel sizing and styling to better adapt across window sizes.

• Fixes: Merged community PRs to squash various UI/function bugs—e.g. resolving the always-expanded settings issue, and preventing GPT text from overlapping interface elements. Marks the 8.x series as mature and stable.

8.0

🛠 IXL Auto Answer – Changelog v8.0
Release Date: 2025-04-12

🔧 Core Updates:

• 🧠 New Model Added: chatgpt-4o-least

  • A human-feedback tuned RLHF variant of gpt-4o
  • High potential, low reliability — more human-like, more prone to mistakes
  • Known for hallucinations, overconfidence, and instability
  • Comes with a warning in the model description — useful for experimentation, not recommended for critical tasks

• ✅ API Key Testing Overhaul

  • No longer relies on external services (e.g. keytest.obanarchy.org)
  • Now uses a real request to the OpenAI /v1/models endpoint
  • Displays full raw server response in a closable popup window
  • Provides better visibility and debugging of API key validity

• 🧠 Improved System Prompt for GPT

  • Explicit instruction to always use the most robust selectors (e.g. XPath)
  • Requires GPT to return complete, standalone JavaScript code
  • Response must be wrapped in proper javascript blocks with no additional commentary

💻 Interface Enhancements:

• 📐 Expanded Control Panel Width: 500px for improved readability
• 📊 Real-time Token Usage Counter displayed in the panel header
• 📤 Rollback Feature retained — allows restoring the previous question state
• 🧪 API Key test results now shown in a scrollable, formatted popup with full JSON response

That‘s it ...

7.3

📌 IXL Auto Answer - Changelog

This document records major changes, enhancements, and fixes in the IXL Auto Answer script.

---

v7.3 - Minor Update 🚀

🆕 New Features:

• 🛠 add something about logging (you now can see the script logs)

---

🛠 Optimizations:

• ⚡ Performance Improvements

  • Faster HTML Extraction: Improved XPath selection for better page structure detection.
  • Smarter Image Processing: The script now skips canvas extraction for text-only problems, reducing unnecessary API requests.

• 🖥️ Enhanced GUI Experience

  • Reorganized control panel with grouped settings (API Key, API Base, Model Selection, Language, Auto Answer, Auto Submit).
  • Panel is now fully draggable and automatically adjusts positioning for better usability.

• 📌 Improved Code Generation

  • More accurate JavaScript execution with better selector detection (id, name, data-*).
  • Regex filtering now prevents unwanted scripts from being executed.

---

🐞 Bug Fixes:

• 🛠 Fixed Auto-Submit Issues:

  • Some users reported that auto-submit was not triggering correctly; this has been fixed.

• 🎯 Improved Answer Detection:

  • The script now ensures all answer fields are properly filled before submission.

• 🔍 Fixed Canvas Image Capture Bug:

  • Previously, some canvas elements were incorrectly captured as blank images. The new logic detects and excludes empty canvases, improving efficiency.

---

🎯 Summary of v7.3

✅ More customization – API Base setting, improved model selection.
✅ Better user feedback – Step-based progress bar, error detection.
✅ Improved accuracy & efficiency – Faster processing, better answer filling.
✅ More stable & reliable – Fixed key bugs, improved UI interaction.

🚀 This update makes IXL Auto Answer more powerful, customizable, and user-friendly than ever before!

7.0 stable version

📌 IXL Auto Answer - Changelog

This document records major changes, enhancements, and fixes in the IXL Auto Answer script.

---

v7.0 - Major Update 🚀

🆕 New Features:

• 🔧 API Base Configuration

  • Users can now set a custom API Base URL in the GUI.
  • This allows the use of alternative OpenAI-compatible APIs instead of the default OpenAI endpoint.

• 📊 Step-Based Progress Bar

  • The script now provides real-time feedback with a dynamic progress bar that updates based on execution steps:
    • ✅ 10% - Started processing
    • ✅ 20% - Extracted question HTML
    • ✅ 40% - Captured canvas images (if applicable)
    • ✅ 40%-95% - Simulated waiting progress while requesting OpenAI API
    • ✅ 100% - Answer processing completed

• 🤖 Expanded Model Selection

  • Added support for the following AI models:
    • GPT-4o - Handles text and image-based problems, cost-effective.
    • GPT-4o-mini - Best for text-only questions, cheapest option.
    • o1 - Highest accuracy for image-based problems, but slow and expensive.
    • o3-mini - Faster than o1, suitable for text-only questions, good balance between cost and accuracy.

• 📌 Improved Error Detection

  • The script now identifies request failures and displays specific error messages:
    • Invalid API Key: Detects incorrect or expired keys.
    • Rate Limit Exceeded: Notifies the user when OpenAI request limits are reached.
    • Network Issues: Alerts users if the API endpoint is unreachable.

---

🛠 Optimizations:

• ⚡ Performance Improvements

  • Faster HTML Extraction: Improved XPath selection for better page structure detection.
  • Smarter Image Processing: The script now skips canvas extraction for text-only problems, reducing unnecessary API requests.

• 🖥️ Enhanced GUI Experience

  • Reorganized control panel with grouped settings (API Key, API Base, Model Selection, Language, Auto Answer, Auto Submit).
  • Panel is now fully draggable and automatically adjusts positioning for better usability.

• 📌 Improved Code Generation

  • More accurate JavaScript execution with better selector detection (id, name, data-*).
  • Regex filtering now prevents unwanted scripts from being executed.

---

🐞 Bug Fixes:

• 🛠 Fixed Auto-Submit Issues:

  • Some users reported that auto-submit was not triggering correctly; this has been fixed.

• 🎯 Improved Answer Detection:

  • The script now ensures all answer fields are properly filled before submission.

• 🔍 Fixed Canvas Image Capture Bug:

  • Previously, some canvas elements were incorrectly captured as blank images. The new logic detects and excludes empty canvases, improving efficiency.

---

🎯 Summary of v7.0

✅ More customization – API Base setting, improved model selection.
✅ Better user feedback – Step-based progress bar, error detection.
✅ Improved accuracy & efficiency – Faster processing, better answer filling.
✅ More stable & reliable – Fixed key bugs, improved UI interaction.

🚀 This update makes IXL Auto Answer more powerful, customizable, and user-friendly than ever before!

6.1 stable version

The IXL is now adaptable for all countries;
bug fix;
license updated;

What's Changed

• Match any domain under ixl.com instead of just ca.ixl.com by @Vortron-rd in #5

New Contributors

• @Vortron-rd made their first contribution in #5

Full Changelog: greasyfork...greasyfork_2024-12-05

5.6 stable version

5.6 stable version
I fixed all bugs i know
https://greasyfork.org/scripts/517259-ixl-auto-answer-openai-api-required?version=1487176