Hugging Face's logo

Datasets:
Offlin33er
/
solidity-exploit-reasoning-v4-sample

Tasks:
Text Generation
Text Classification
Modalities:
Text
Formats:
csv
Languages:
English
Size:
< 1K
Tags:
solidity
smart-contracts
ai-agents
mcp-security
llm-alignment
code-auditing
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
solidity-exploit-reasoning-v4-sample / README.md
Offlin33er's picture
Offlin33er
Update Oxford Digital v4 commercial dataset card ($149 Pro / $499 Enterprise)
41f5a76 verified
preview code
|
Raw History Blame Contribute Delete
4.92 kB
metadata 
license: other
license_name: oxford-digital-commercial-data-license-1.0
license_link: https://oxforddigitals.com/data-license
task_categories:
  - text-generation
  - text-classification
language:
  - en
tags:
  - solidity
  - smart-contracts
  - ai-agents
  - mcp-security
  - llm-alignment
  - code-auditing
  - agent-benchmark
pretty_name: Solidity & AI Agent Security Remediation Benchmark v4 (100-Row Preview)
size_categories:
  - 1K<n<10K

Solidity & AI Agent Security Remediation Benchmark v4 — Commercial Preview

Curated by Oxford Digital (Offlin33er)

Instant Commercial Access — Full 1,600-Row Archive (Parquet + CSV + JSONL)

This Hugging Face repository hosts a free 100-row stratified preview so ML engineers, AppSec teams, and Web3 auditors can inspect exact schemas and remediation quality directly in the Dataset Viewer above.

What Is Inside the Full v4 Archive (1,600 Verified Records)

  1. Solidity v0.8.20–0.8.26 Smart Contract Audit & Remediation Pairs (800 Records):
    • Tokenized vault share accounting & virtual offsets, async settlement state ordering, two-step governance controls, oracle staleness validation, EIP-712 domain/nonce replay protection, proxy initializer lockouts, bounded pull-claim distributions, and fee-on-transfer balance-delta accounting.
  2. AI Agent, MCP, RAG & LLM Guardrail Remediation Pairs (800 Records):
    • Python & TypeScript implementations covering MCP tool workspace confinement, RAG untrusted-context isolation, autonomous agent loop/budget circuit breakers, strict Pydantic structured output parsing, human-in-the-loop cryptographic approval gates, read-only parameterized Text-to-SQL tooling, multi-tenant vector ACL filtering, and deterministic RLVR reward graders.
  3. Zero Public Overlap:
    • 100% private, deduplicated records cryptographically verified (0 hash collisions) against public Offlin33er/solidity-audit-instruct (v1–v3) splits to prevent benchmark contamination.

Pricing Tiers

Tier Price License Scope
Self-Serve Pro $149 USD Perpetual commercial AI training, fine-tuning (SFT/DPO/GRPO/RLVR) & evaluation for a single engineering team (up to 10 seats).
Enterprise $499 USD Organization-wide perpetual commercial license across unlimited internal teams & subsidiaries + priority v4.x updates (Card, Invoice, or USDC via rayxxxnd@oxforddigitals.com).

Dataset Schema

Column Type Description
id string Deterministic SHA-256 derived record identifier (oxd_...)
category string Normalized security domain (Access & Trust Boundary Control, State & Execution Integrity, Resource & Economic Safety)
severity string Normalized risk rating (Critical, High, Medium)
input_artifact string Unhardened Solidity smart contract or Python/TypeScript AI agent/MCP/RAG implementation
audit_reasoning string Architectural root-cause analysis and violated security invariant
remediated_output string Hardened, production-ready Solidity, Python, or TypeScript implementation

Related Open-Source Models by Oxford Digital (Offlin33er)

Commercial License Summary

The 100-row sample in this repository may be inspected freely for schema and quality evaluation. Full dataset archives licensed via Oxford Digital grant perpetual internal rights to train, fine-tune, and evaluate proprietary or commercial AI models, while prohibiting raw dataset redistribution or resale. Full terms: oxforddigitals.com/data-license.