Datasetpublic

nisaefendioglu/synthetic-sensitive-data-in-source-code-n300

Synthetic Sensitive Data in Source Code (N=300) Synthetic dataset of 300 source-code / config snippets containing hardcoded secrets and PII.Every sample includes at least one sensitive finding (no clean negatives in the main file). Designed for local masking, secret detection, and OWASP LLM02 — Sensitive Information Disclosure workflows. Version 1.3.6: README Files table documents split extension (split_pattern_custom_n300) and clean CSV. v1.3.4: split files + Dataset Viewer… See the full description on the dataset page: https://huggingface.co/datasets/nisaefendioglu/synthetic-sensitive-data-in-source-code-n300.

sourceHugging Facemitupdated 12d agoView on Hugging Face

1likes170downloads

Dataset Card

Synthetic Sensitive Data in Source Code (N=300)

Synthetic dataset of 300 source-code / config snippets containing hardcoded secrets and PII. Every sample includes at least one sensitive finding (no clean negatives in the main file).

Designed for local masking, secret detection, and OWASP LLM02 — Sensitive Information Disclosure workflows.

Version 1.3.6: README Files table documents split extension (split_pattern_custom_n300) and clean CSV. v1.3.4: split files + Dataset Viewer config split_pattern_n300. v1.3.2: clean extension (clean_n300).

All values are synthetic / fake. Do not treat them as real credentials.

Dataset Viewer

Use configs `sensitive_n300` (default), `clean_n300`, and `split_pattern_n300` for CSV previews. Download JSON for full sensitive_findings / custom_rules_recommended.

Files

File	Description
`synthetic_sensitive_data_in_source_code_n300.json`	Full records + ground truth
`synthetic_sensitive_data_in_source_code_n300.csv`	Flat view
`starter_explore_sensitive_code.ipynb`	Starter notebook
`extensions/clean_code_snippets_n300.json`	Clean negatives (N=300)
`extensions/clean_code_snippets_n300.csv`	Clean negatives — flat view
`extensions/split_pattern_custom_n300.json`	Split / custom-regex positives (N=300) + `custom_rules_recommended`
`extensions/split_pattern_custom_n300.csv`	Split extension — flat view (Viewer config `split_pattern_n300`)
`metadata/dataset_stats.json`	Stats (not a split)

Extension: clean negatives (N=300)

File	Description
`extensions/clean_code_snippets_n300.json`	300 snippets without labeled secrets
`extensions/clean_code_snippets_n300.csv`	CSV export

Extension: split / indirect secrets (N=300)

File	Description
`extensions/split_pattern_custom_n300.json`	Full records + `custom_rules_recommended`
`extensions/split_pattern_custom_n300.csv`	Flat view for Dataset Viewer

For evaluating built-in vs custom-regex masking on concatenated, folded, or comment-embedded secrets (thesis extension set).

Citation

Efendioğlu, N. N. (2026). Related thesis / dataset publication on Hugging Face (DOI: 10.57967/hf/9800).