phi

Self-evolving recursive AI reasoning system — φ-governed tool use, multi-agent code review, workflow orchestration, and persistent spiral memory, built on E8 Lie group geometry and golden ratio mathematics

HIPAA Safe Harbor compliant DICOM de-identification — Python library, CLI, and REST API. Removes 18 PHI identifiers, remaps UIDs, FastAPI service.

LangChain wrapper for Phi-4 models via Ollama.

PHI-shaped redaction layer that must clear a recall gate before it may sit in front of an LLM

MCP server for redacting PII/PHI and de-identifying healthcare text before it reaches an LLM. self-hosted, fail-closed, HIPAA-aware.

Local-first hybrid clinical-text de-identification platform with contextual ML, human-review controls, and FHIR export.

The Parma Health Toolkit is designed as a modular, local-first Python library that sits between raw sensitive data sources and downstream analytics or cloud environments. It functions as a "sanitization gateway," ensuring no Protected Health Information (PHI) leaves the secure source zone without anonymization.

Context-aware PII/PHI detection & masking for healthcare data: a stateless OSS engine (on PyPI), a hardened FastAPI service (JWT RBAC + encryption), and a Next.js app. Built on Microsoft Presidio + spaCy, model-configurable (sm/md/lg/trf).

Prove your PHI de-identification actually works — a CI gate that runs a synthetic HIPAA Safe Harbor corpus through your de-id and fails the build if any of the 18 identifiers survive. Never touches real PHI.

Alpha · Tier 1 — PHI de-identification + time-series anomaly detection + compliance checks + MCP bridge. phantom-mesh security suite.

Local proxy that removes patient and client details from AI prompts before they reach OpenAI or Anthropic, then puts them back in the reply. Built for HIPAA and GDPR. Early development.

Local AI infrastructure for multi-LLM RAG, vector search, and agentic workflows. Combines Ollama, FastAPI, ChromaDB, and LangGraph for production-style open-source LLM orchestration and retrieval-augmented generation.

Best-effort PHI redaction CLI with deterministic detection, model review, and evaluations.

HIPAA compliance skills for AI coding agents (Claude Code, Codex, Cursor, Gemini). 10 senior-grade skills: scoping, app dev, websites, AI/LLM, code review + PHI scanner, risk analysis, breach response, BAAs, de-identification, compliance programs. Exact 45 CFR citations, OCR enforcement through 2026.

A local tool that automatically detects and bleeps out Protected Health Information (PHI) from audio files.

HIPAA-compliant PHI scrubber — blocks CDT/CPT/ICD-10 codes, tooth numbers, diagnosis terms before they leave your server. Extracted from usdpr. (US Dental Patient Recovery).

Local clinical text de-identification with reversible, deterministic PHI tokens. Strip PHI before it leaves your machine; restore responses locally.

ChatGPT interface with custom UI

A simple implementation of Euler's totient function ("Phi function").

MCP server and CLI for detecting, redacting, and auditing PHI before medical text is sent to AI agents.