Krishnapadala55/Brahmastra-DarkNetra

---
license: cc-by-nc-sa-4.0
language:
- en
tags:
- security
- cybersecurity
- dast
- vulnerability
- penetration-testing
- owasp
- web-security
- appsec
- red-team
size_categories:
- 100K<n<1M
task_categories:
- text-generation
pretty_name: "Brahmastra DarkNetra"
dataset_info:
features:
- name: messages
dtype: string
- name: source
dtype: string
- name: license
dtype: string
- name: category
dtype: string
- name: turns
dtype: int32
---

# Brahmastra DarkNetra

*The Dark Eye that sees every vulnerability in the shadows.*

> **Security Research Dataset Notice**: This dataset contains cybersecurity training data
> including descriptions of vulnerability exploitation techniques, security testing payloads,
> and attack methodologies for educational and defensive purposes. Antivirus software may
> flag files due to pattern matching on security-related text. This is expected behavior
> for cybersecurity datasets and the files do NOT contain executable malware. Raw exploit
> payloads have been sanitized to reduce false positives while preserving educational value.

The largest open-source ChatML dataset for training AI-powered security scanners and cybersecurity assistants. Named from Sanskrit: Netra = eye, vision. DarkNetra sees what others miss.

## Dataset Summary

| Metric | Value |
|--------|-------|
| **Total samples** | **838,679** |
| Train split | 813,519 |
| Test split | 25,160 |
| Format | ChatML (messages array) |
| License | CC BY-NC-SA 4.0 |

## What's Inside

| Category | Samples | Description |
|----------|---------|-------------|
| cve-knowledge | 346,212 | |
| pentest-methodology | 318,809 | |
| exploit-knowledge | 94,870 | |
| synthetic-dast | 49,500 | |
| cybersec-reasoning | 31,853 | |
| bugbounty | 10,432 | |
| scanner-intelligence | 10,261 | |
| attack-methodology | 6,590 | |
| tool-knowledge | 5,784 | |
| attack-knowledge | 4,426 | |
| owasp-standards | 3,976 | |
| lab-writeups | 1,203 | |

## Sources

| Source | License | Samples |
|--------|---------|---------|
| Pentest-Agent | Apache-2.0 | 318,809 |
| All-CVE-Records | CC-BY-4.0 | 297,346 |
| BRAHMASTRA-Synthetic | MIT | 49,500 |
| NVD-Web-CVEs | CC-BY-4.0 | 48,866 |
| ExploitDB | GPL-2.0 | 48,048 |
| ExploitDB-WAF | MIT | 46,822 |
| Cybersec-32K | Apache-2.0 | 31,853 |
| BugBounty-Public | CC-BY-NC-SA-4.0 | 10,432 |
| Scanner-OSS | Mixed-OSS | 10,261 |
| HackTricks | CC-BY-NC-4.0 | 6,590 |
| Tool-Intelligence | MIT | 5,784 |
| PayloadsAllTheThings | MIT | 4,426 |
| OWASP-Top10 | CC-BY-SA-4.0 | 1,594 |
| PortSwigger-Community | CC-BY-NC-SA-4.0 | 1,203 |
| OWASP-WSTG | CC-BY-SA-4.0 | 998 |
| OWASP-MAS | CC-BY-SA-4.0 | 813 |
| OWASP-ASVS | CC-BY-SA-4.0 | 382 |
| OWASP-SAMM | CC-BY-SA-4.0 | 189 |

## Sample Format

```json
{
"messages": [
{"role": "system", "content": "You are a cybersecurity expert..."},
{"role": "user", "content": "How do you test for blind SQL injection?"},
{"role": "assistant", "content": "<think>Blind SQLi requires...</think>..."}
],
"_source": "BRAHMASTRA-Synthetic",
"_license": "MIT",
"_category": "synthetic-dast",
"_turns": 3
}
```

## Usage

```python
from datasets import load_dataset
ds = load_dataset("Krishnapadala55/Brahmastra-DarkNetra")
print(f"Train: {len(ds['train']):,} | Test: {len(ds['test']):,}")
```

## License

CC BY-NC-SA 4.0 - See individual `_license` fields per sample for source-specific terms.

## Citation

```bibtex
@dataset{brahmastra_darknetra_2026,
title={Brahmastra DarkNetra: CyberSec DAST Dataset},
author={Krishna Padala},
year={2026},
url={https://huggingface.co/datasets/Krishnapadala55/Brahmastra-DarkNetra}
}
```

Built with [BRAHMASTRA](https://github.com/brahmastra-dast/brahmastra) - *Precision. Power. It never misses.*