Oriel — AI & technology worth noticing, every day.

PentesterFlow/OffsecCyber

Dataset · https://huggingface.co/datasets/PentesterFlow/OffsecCyber

Security Vulnerability Instruction Dataset An English-language dataset of security-focused assistant conversations about software vulnerabilities. Examples cover vulnerability explanations, CVE analysis, severity and impact, CWE references, and remediation guidance. It combines CVE-oriented examples with HackerOne report conversations and can support instruction tuning or experiments in security-topic classification and analysis. What the data contains Each… See the full description on the dataset page: https://huggingface.co/datasets/PentesterFlow/OffsecCyber.

First seen by Oriel on .

AI summary

PentesterFlow/OffsecCyber is a curated 60,517-record web and application security conversation dataset for instruction tuning, vulnerability analysis, and security-topic classification.

  • It combines synthetic CVE analysis, NVD records, offensive web-appsec examples, HackerOne reports, PortSwigger labs, and a small set of tool-use and negative examples.
  • The cleaned corpus reduces near-duplicate redundancy from 40% to 6%, raises class-label coverage to 95.3%, and diversifies the dominant system prompt from 57% to 9%.
  • Train, validation, and test splits use a seeded 90/5/5 ratio and group shared CVE IDs, HackerOne report IDs, and exact conversation hashes to reduce leakage.
  • The dataset is web-appsec focused: about 0.7% of records concern memory safety and only about 3% of answers contain code, so it is not intended for binary PoC reproduction.
  • Labels and vulnerability details may be synthetic or incomplete; class frequencies are imbalanced, and the dataset includes no benchmark results.

#cybersecurity #web security #instruction tuning #vulnerability classification

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Datasets: 109, latest observation ; +98 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Datasets: 8,628, latest observation ; +8,628 over observed period since

Aliases

SourceExternal IDURL
Hugging Face DatasetsPentesterFlow/OffsecCyberhttps://huggingface.co/datasets/PentesterFlow/OffsecCyber Read discussion