SWE Multi-Domain Demo Pack — 10 Executable Software-Repair Worlds | Synthetic Data | Worldwide
Overview
SWE Multi-Domain Demo Pack — 10 Executable Software-Repair Worlds | Synthetic Data | Worldwide is a text dataset. Typical applications include evaluation & benchmarking, document understanding & OCR and agents & tool use. Relevant sectors: AI & Machine Learning and Education. The listing mentions classification labels. Pricing is not published and is provided on request.
Primary applications
What the listing offers
- Listing mentions classification labels.
A compact, execution-validated synthetic dataset for evaluating software-repair agents: 10 fictional microservices, one per industry, each containing exactly one labelled defect with a verified fix. THE 10 INDUSTRIES — AND THE DEFECT EACH WORLD CONTAINS 1. Fintech (CWE-190, Integer Overflow) — payment-amount arithmetic overflows at scale, silently accepting or rejecting transactions at boundary values. 2. E-commerce (CWE-129, Improper Array Bounds) — checkout reads a cart index that a concurrent stock re-sync can invalidate, crashing purchases mid-checkout. 3. DevOps / Infrastructure (CWE-703, Improper Check) — a service silently ignores TLS certificate expiry, failing closed only long after the certificate lapsed. 4. Healthcare (CWE-668, Resource Exposure) — an ambulance GPS service serves stale location data during vehicle handover, misdirecting dispatch. 5. Logistics / Freight (CWE-681, Incorrect Conversion) — manifest weights are truncated to whole numbers, so bills of lading disagree with recorded pallet weights. 6. Gaming backend (CWE-20, Improper Input Validation) — an in-game store accepts negative quantities, turning purchases into an unlimited currency mint. 7. EdTech (CWE-697, Incorrect Comparison) — students exactly on the pass mark are reported as failed because the boundary check is exclusive. 8. Travel / Booking (CWE-682, Incorrect Calculation) — seat availability over-counts held reservations, overselling flights. 9. Energy / Grid (CWE-190, Integer Overflow) —
Related datasets
More in Multimodal & Agent Training →Prescription, EMR & Clinical Records Dataset - 1.06M+ Patients, 976K+ Documents, 7 Clinical Categories | Clinical Data
Prescriptions, EMR & Clinical Records Dataset The Prescriptions, EMR & Clinical Records dataset is a comprehensive collection of anonymized electronic medical records…
- Language
- —
- Volume
- 976k documents
- Domain
- Healthcare +2
- Use case
- Evaluation & benchmarking