Matex Privacy Sentinel: Synthetic Data for Training Local Privacy Filters

A synthetic token-classification dataset created by enosislabs for training a local privacy and security filter. The dataset is designed to teach a filter to detect and redact sensitive information spans in developer workflows before data is sent to external large language models. It was last updated on Hugging Face on 2026 05 02.

Use Cases

• Train a local filter to redact sensitive data from .env files based on the description.
• Develop a model to sanitize terminal logs and stack traces before external transmission.
• Create a privacy filter for git diffs and GitHub issue comments as mentioned in the description.
• Secure agent traces, tool outputs, and workspace memory by detecting sensitive spans.
• Filter authentication, database, cloud, and payment information from developer contexts.

Strengths

• Dataset is specifically designed for a concrete purpose: training a local privacy filter for developer workflows.
• Targets a defined set of sensitive domains including .env files, terminal logs, and git diffs as listed in the description.
• Last updated date of 2026 05 02 22:37:54 is provided.

Limitations

• Row count, file formats, and column-level documentation are unknown, which may limit suitability assessment.
• The dataset is synthetic; its effectiveness on real-world data requires validation.
• License information is unknown, which may restrict usage.

Provenance

Source

enosislabs on Hugging Face.

Collection Method

Synthetically generated, likely based on the OpenAI Privacy Filter.

Time Range

null

Freshness

Last updated 2026 05 02 22:37:54.

Geography

null