Tula is an open-source collection of OpenClaw skills, configurations, and patterns designed to transform a general-purpose AI agent into a personal health intelligence assistant.

NoHarm API for Smart Prescription Screening

MedRiskEval: Medical Risk Evaluation Benchmark of Language Models, On the Importance of User Perspectives in Healthcare Settings (EACL 2026 Industry Track)

AI system for detecting look-alike sound-alike (LASA) drug name errors using machine learning, NLP, and clinical context validation.

Find patients who have concerning tests but no timely follow-up

A web-based visualization tool for patient-oriented anesthesia risk visualisation

Clinician-built benchmark and live leaderboard for medical AI safety evaluation.

Open-source evaluation framework for healthcare AI applications.

Source code for patientaider.org

A patient-safety-focused desktop AI assistant providing structured first-aid guidance using a hybrid RAG pipeline with medical-specialized embeddings and Google Gemini.

Continuity-aware clinical governance infrastructure for autonomous healthcare systems.

Clinical simulation intelligence for AI-assisted reports, Latent Safety Threat tracking, and institutional learning at WashU Emergency Medicine.

A rules-first AI medical triage and lab-result interpretation web application

IoMT risk education platform built from 28 years of clinical emergency medicine and lived device dependency. Real failure scenarios. No hypotheticals.

AI-powered clinical safety netting for NHS primary care. Red flag detection, automated patient follow-up, GP escalation.

Скрипт анализирует базу данных пациентов и используется заведующим отделением пластической хирургии и главным врачом для контроля пациентов с инфекционными заболеваниями при планировании операций

Physician-curated safety prompts, guardrails, and red-team benchmarks for healthcare LLMs. Built by a doctor for AI teams shipping in healthcare.

ECG AI that measures how dangerous its mistakes are, not just how often it's wrong.

LLM evaluation harness for nursing-workflow tasks, built by a working cardiac nurse — blinded clinician scoring, fabrication penalties, and a clinician-review gate enforced in code

Reproducible paired benchmark of evaluator-dependent conclusions in audience-calibrated medical LLM safety assessment