← All sample reportsSynthetic sample code only (F18) — no real student or candidate work.
ai-code-detector · screening reportpage 1 / 4
Screening report · sample
take-home-2026-09 — candidate batch screening
Job ID
demo_job_takehome
Submitted
2026-09-10 09:05 UTC
Files
4 files · 512 lines total · JavaScript (4)
Engine version
codebert-ft-0.1.0
Retention
Sample report — not stored, nothing to delete
NoteThis score estimates the likelihood that code was AI-generated. It is a screening signal, not proof of AI use. False positives and false negatives occur. Always review flagged work with the author and follow your institution’s or company’s policies before taking any action.
ai-code-detector · screening report demo_job_takehome · screening signal, not proof — review flagged work with its author
Sheet 2 · Summary — per-file scores, sorted
summarypage 2 / 4
Per-file scores, highest first
Start your review at the top of this table. A high score means "review this first" — nothing more.
#
File
Lines
AI-likelihood score
Flagged lines
1
rate_limiter.js
141
74
30–38
2
api_client.js
96
61
12–15
3
retry_queue.js
158
33
—
4
config.js
117
12
—
0–6 · lower signal
7–13 · mixed signal
14–100 · stronger signal
ai-code-detector · screening report demo_job_takehome · screening signal, not proof — review flagged work with its author
Sheet 3 · File detail — side-by-side evidence
file detail · rate_limiter.jspage 3 / 4
rate_limiter.js
74AI-likelihood score
Flagged code
30async function processRequestQueue(queue) {
31 // Process each request in the queue sequentially
32 const results = [];
33 for (const request of queue) {
34 // Execute the request and store the result
35 const result = await executeRequest(request);
36 results.push(result);
Why these segments were flagged
lines 30–35
Every step carries a comment that restates the next line — a generator-style pattern worth a follow-up question.
overall
Naming and error-handling style are uniform across branches. Ask the candidate to walk through the backpressure choice.
Signal groups referenced: naming patterns · comment density · structural predictability · edge-case handling.
ai-code-detector · screening report demo_job_takehome · screening signal, not proof — review flagged work with its author
Sheet 4 · How to use this report + archival record
how to use this reportpage 4 / 4
How to use this report
This report is a record of a screening step. It is not a finding of misconduct, and it is not a basis for a grade penalty or a hiring decision on its own.
Ask the author to explain the flagged snippets — people who wrote the code can usually walk through it.
Ask them to extend or modify the flagged code on the spot, with a small change in requirements.
Compare with the author's earlier work before drawing any conclusion.
NoteThis score estimates the likelihood that code was AI-generated. It is a screening signal, not proof of AI use. False positives and false negatives occur. Always review flagged work with the author and follow your institution’s or company’s policies before taking any action.
ai-code-detector · screening report demo_job_takehome · screening signal, not proof — review flagged work with its author