wardcat

Overview

On-prem detection and anonymization of PII and sensitive data in LLM inputs. Combines regex, named entity recognition, and a local LLM to identify and mask sensitive information. Written in Python as a community project.

Best for

Best for
Developers needing lightweight on-prem PII redaction for LLM workflows.

Use cases

• Anonymize PII in prompts before sending to external LLMs
• Detect sensitive data in text corpora
• Build privacy-preserving LLM pipelines on local infrastructure

Notes

On-prem detection and anonymization of PII and sensitive data in LLM inputs. Combines regex, named entity recognition, and a local LLM to identify and mask sensitive information. Written in Python as a community project.

2 stars on GitHub. Last updated 2026-10-04. Licensed MIT.

Use cases

• Anonymize PII in prompts before sending to external LLMs
• Detect sensitive data in text corpora
• Build privacy-preserving LLM pipelines on local infrastructure

Pros

• Runs on-prem, keeping data in-house
• Hybrid approach improves detection coverage
• Open source and free to use

Cons

• Early-stage project with limited community adoption (2 stars)
• Requires local LLM setup and maintenance
• May need tuning for domain-specific PII

Indexed from awesome-llmops and enriched against its public facts.