| name | test-taker-be-corrective |
| description | Take the backend corrective (debugging) capability benchmark - create a submission folder named <llm>+<harness> under submissions/be-corrective/ and fix all nine broken Node backends (api/db/service × easy/mid/hard). Use when asked to take the be-corrective test, sit the backend debugging benchmark, or generate a be-corrective submission. |
| disable-model-invocation | true |
Test Taker (BE Corrective)
You are sitting the be-corrective (backend debugging) capability benchmark. Each question gives you mostly-correct Node.js code with injected bugs. Fix them. Work independently and do your best.
Rules (read before starting)
- Read only from
questions/be-corrective/<question>/. Never open, read, or reference reference-answers/ - consulting it invalidates the run.
- Write only inside your own submission folder. Do not modify
questions/, other submissions, or anything else in the repo.
- stdlib Node only — no Express, no SQLite/Redis, no Docker, no new npm dependencies. Keep the same filenames.
- Budget roughly 6 minutes per question. Higher difficulty = more and subtler bugs.
Step 1 - Create your submission folder
Name it <llm>+<harness> in lowercase with no spaces.
submissions/be-corrective/<llm>+<harness>/
q1-api-easy/ q1-api-mid/ q1-api-hard/
q2-db-easy/ q2-db-mid/ q2-db-hard/
q3-service-easy/ q3-service-mid/ q3-service-hard/
Step 2 - Answer each question
Read each PROMPT.md and the broken sources, then write the fixed files into the matching submission subfolder (same filenames). Copy data files if present (data/*.json) unless the prompt says they are read-only fixtures you should not change — prefer fixing code, not fixtures.
| Question | Typical files to fix | What you are doing |
|---|
q1-api-* | server.js (+ keep package.json) | HTTP CRUD / pagination / reservations |
q2-db-* | lib/queries.js or lib/analytics.js or lib/cohort.js (+ data/*.json fixtures) | JSON “DB” query/aggregate bugs |
q3-service-* | server.js or lib/ledger.js | auth / sessions / concurrent transfers |
Constraints
- Export whatever the prompt specifies (
createServer, reset, query functions, etc.) so the checker can import or fetch.
- Scoring is binary: 10 if the checker PASSes, else 0.
Step 3 - Finish
Verify all nine submission subfolders have their fixed sources, then report your submission folder name. Do not grade yourself.
