Seeking Feedback: Open Source (GPL-3.0) AI Toolkit for 2026 ADA Math & Table Remediation

Question

Hello Accessibility Community,

I am an instructor of Software Application Design and Development at McHenry County College with 26 years of teaching experience. As we approach the April 2026 ADA deadline, I am working on an open-source project to help automate the remediation of complex Canvas content.

I am looking for community members to help test the accuracy of a free, copyleft tool I am building called MOSH (Making Online Spaces Helpful).

Full AI Disclosure:

* The Tool: MOSH integrates Gemini 2.0 Flash to assist in converting PDF/Handwritten math into LaTeX and OCRing complex tables into semantic HTML. It includes a "Human-in-the-Loop" Visual Review Dashboard to ensure AI outputs are verified by an instructor.
* This Post: I have utilized AI assistance to help draft and refine this communication to ensure it meets community standards and clearly explains the technical scope.

How to Help:Because I am not a mathematician, I specifically need "Math Eyes" to verify if the AI-generated LaTeX is rendering accurately for higher-level STEM courses.

* Download: Grab the latest release candidate from the MOSH GitHub Repository.
* Test: Run a few "problem" math PDFs or handwritten notes through the toolkit.
* Feedback: Let me know if the AI is accurately capturing notations or if it is "hallucinating" results.

Is anyone else currently utilizing AI-orchestration for the 2026 deadline? I would love to hear your feedback on this approach.

Best,

Meredith Kasprak, PhDInstructor, McHenry County CollegeLead Developer, MOSH Toolkit

Meri_Kasprak · Answer

Yes, they will have to acknowledge that it doesn't currently have a security certificate. I need to apply for one. Windows will warn that it is a potential danger. I strongly recommend putting it into an empty course shell, then verifying the content as you move it into use. This is only two months into development. It does a great job on many things, but I need to test with many different classes and teachers to be confident before I consider it a released version. Math, I'm still working on, but it's promising, and I hope to have a testable version next week.

bwsf · Answer

Thanks for sharing this! I'm talking with our math department tomorrow about accessibility, and I'll see if anyone is up for beta-testing this.

I just watched your intro video (love the Mosh duck!).

It looks like I should tell people who are interested:

* Ask us for an empty course shell to put processed files into.
* It currently works on Windows, and you just have to download the .exe file and open it. (my reading of the readme)
* This is under development; please understand that and provide feedback if you can.

Do they have to install or otherwise configure Python to get this to work? (Or does this cause them problems if Python is already installed?)

Thank you!

Meri_Kasprak · Answer

I wanted to clarify that while I am seeking "Math Eyes," MOSH is a comprehensive remediation suite for all Canvas courses.

For "Normal" (Non-Math) Classes:The tool includes a robust Non-AI Component that programmatically scans and fixes common WCAG errors. This includes structural cleanup (fixing broken HTML nesting) and identifying "click here" style links that need descriptive text.

How it handles Images and Links:One of the core features of the Visual Review Dashboard is keeping the instructor in the driver's seat:

* Image Context: When the tool identifies an image, it "pops up" for the user. The image is visible with a text entry box below it so all images in the course can quickly get alt tags. 
* Link Verification: Similarly, for links, the tool shows the surrounding text and allows the user to click the link to verify the destination before updating the link text to be more descriptive.

AI-Enhanced Responsive Design:When utilizing the Gemini integration, MOSH goes beyond simple fixes to modernize legacy content. It converts old, fixed-width tables into responsive HTML/CSS, ensuring that content scales correctly for students using the Canvas Student App on mobile devices.

Code & Feedback:The project is open-source (GPL-3.0) and built with Python (BeautifulSoup4 and Tkinter). I’ve just updated the MOSH GitHub Repository, and I would welcome feedback on the structural HTML logic, the UI workflow, or the math parsing!