🏥 OpenClaw Medical Harness

Medical AI Agent Orchestration Framework — Built on Harness Theory

📦 PyPI v0.2.0 ⭐ GitHub ✅ CI 🌐 Ecosystem

Quick Start

pip install openclaw-medical-harness

from openclaw_medical_harness import DiagnosisHarness
h = DiagnosisHarness(specialty='neurology')
r = h.execute({'symptoms': ['ptosis', 'fatigable weakness'], 'patient': {'age': 35}})
print(r['diagnosis'], r['confidence'])

Three Harnesses

Harness	Pipeline	Integrations
Diagnosis	Symptoms → Differential → Workup → Dx	PubMed, OMIM, Knowledge Base
Drug Discovery	Target → Screening → ADMET → Lead Opt	ChEMBL, OpenTargets, RDKit
Health Mgmt	Assessment → Plan → Adherence → Effect	Wearables, Labs, PubMed

5-Step Pipeline

Input → Context Build → Tool Chain → Model Reasoning → Validation → Recovery → Output

Any LLM (MIMO, Claude, GPT-4, Ollama) + MCP tool chain + medical-grade validation

Model Providers

Provider	Model	Status
Xiaomi MIMO	mimo-v2-pro	✅ Default
OpenAI	GPT-4	🔌 Via factory
Anthropic	Claude	🔌 Via factory
Ollama	Local models	🔌 Via factory

MCP Tools

Tool	Category	Auth
PubMed	Literature	No
ChEMBL	Drug data	No
OpenTargets	Target-disease	No
OMIM	Genetics	API Key
OpenFDA	Drug safety	No

Harness Theory

在AI应用中，架构设计（Harness）比底层模型更重要。完整Harness可将准确率从72.3%提升至91.8%。

MoKangMedical Ecosystem

慢康智枢

Hospital chronic disease management

MediChat-RD

Multi-agent rare disease diagnosis

MediPharma

AI drug discovery platform

DrugMind

Drug R&D digital twin

云思园

AI memorial platform

数字智者

100+ historical AI dialogues

天眼

Multi-agent market forecasting

🌐 Ecosystem Hub

All projects in one place