A Chrome extension for Moodle assignment grading pages. Version 0.1.71 reads supported submissions and task references, combines course, task and case-specific teacher criteria, and requests a suggestion from your chosen AI provider. Review the suggestion before applying it to Moodle's grading fields. It never automatically submits the grading form.
What it does
- Prepare suggestions for rubrics, marking guides or simple direct grading, with the detected mode shown in the panel.
- Read supported PDF, DOCX, ODT and source-code evidence; code is treated as text and never executed. Images require a compatible vision model.
- Let the teacher review suggestions before applying them to Moodle fields and saving in Moodle.
What you need
- Chrome on a computer, manual extension installation, and permission to grade the Moodle assignment.
- A suitable model in Ollama, LM Studio, Jan, GPT4All, llama.cpp, LocalAI, vLLM or an approved OpenAI-compatible API.
- A vision-capable provider/model for scanned pages or images.
Installation
Install the public extension manually
Download
moodlia-corrector-0.1.71.zipfrom the release assets and extract it into a permanent folder. Do not use the generic source-code ZIP.Open
chrome://extensions, enable Developer mode, and choose Load unpacked. Select the extracted folder containingmanifest.json.Open a Moodle assignment grader and click the extension icon. In the correction sidebar, open AI configuration and select the provider and model. Advanced settings include the address, port and other connection options.
Use Test connection with fictional data before reading student submissions. Try the workflow in a test assignment first.
First use
Open an assignment grading page and inspect the extracted submission and assessment criteria.
Request a suggestion, check the evidence and marks, and edit the result as needed.
Apply only the reviewed result to the form, then use Moodle's own save action when you are satisfied.
Good to know
- This is a manually installed extension, not a Chrome Web Store installation. Check the release page for updates.
- Assignment evidence reaches your selected AI provider. Identifier minimization does not make student work anonymous; use only an authorized provider.
- Task references and student submissions are listed separately. Supported files are read when requesting correction; unsupported files are ignored and marked red. Verify which evidence was actually read.
- AI output can be wrong. The teacher remains responsible for reviewing the proposed marks and feedback.