I built a written pre-screening tool for recruiters. Tell me where it breaks

I've made this tool called Eneya, and unlike most things posted here the name tells you nothing, so: it's written pre-screening for tech recruiters. You build a short questionnaire around the competencies the role actually needs, candidates answer it in text on their own time, and you get back a scored report on each one. No video, no scheduling, no calls that end in the first four minutes.

Anyone who has screened for a role with two hundred applicants knows the specific dread of it: you're going to interview maybe eight of those people, the CV tells you almost nothing about who's worth the slot, and somewhere in that pile is the person you should have hired. With Eneya you get to see how each candidate actually thinks about the work before you spend an hour on them, and the report shows its own arithmetic, so when you tell a hiring manager why someone didn't make the shortlist you have something to point at rather than a number the tool refuses to explain.

It's early, real recruiters are using it and breaking it in ways I didn't predict, which is exactly what I need right now. Link is in my profile if you want to run a real role through it and tell me where it falls apart.