Blender File Checker for E-Learning Platform

We're looking to build a Blender file checker for our e-learning platform. If this mission is carried out well, we would like to enter into a long-term collaboration with you. The goal is to design a system capable of automatically or semi-automatically analyzing .blend files submitted by learners, in order to identify errors, inconsistencies, or missing elements in their projects. The mission is not about manually correcting a few files, but about proposing a reusable correction method that can be applied across multiple Blender exercises. You are free to choose whichever approach you find most relevant: a Python script, automation within Blender, AI-based analysis, a combination of code + AI, or any other suitable solution. Goal of the checker Ideally, the checker should be able to: Analyze a Blender .blend file. Verify whether the project meets the general expectations of an exercise. Detect significant errors or inconsistencies. Identify elements that are missing, poorly named, poorly organized, or misconfigured. Generate clear, actionable feedback. Distinguish between what can be corrected automatically and what should simply be flagged to the learner. Intentionally open scope We are deliberately not providing an overly detailed procedure. The checker can rely on various criteria depending on what you judge relevant: Scene structure Objects present or absent Organization of collections Object naming Materials Scale and proportions Positioning of elements Overall file quality Consistency with the expected exercise Visible or detectable technical errors It's up to you to define the most appropriate control logic and the most reliable way to process the files. Expected deliverables At the end of the mission, we expect: The checker, or a prototype of it, so that we can test it. An explanation of how it works. The verification criteria used. The elements needed to understand or verify your approach: code, automation, bot, workflow, or any other system used. The current limitations of the system. Possible avenues for improvement. The aim is to assess your ability to design an autonomous, reusable solution that is reliable enough to correct or analyze Blender project submissions on our e-learning platform.