clinicalguard

Constitutional AI governance for healthcare agents.

clinicalguard is a Starlette-based A2A service for validating proposed clinical actions against a healthcare constitution. It combines deterministic ACGS validation, healthcare-specific validators, JSON-RPC 2.0 endpoints, and audit-trail persistence.

Installation

clinicalguard supports Python 3.11+.

pip install acgs-lite clinicalguard
pip install clinicalguard[anthropic]
pip install clinicalguard[openai]

clinicalguard currently relies on acgs-lite at runtime; install it alongside the package when working from PyPI or source checkouts.

Quick Start

Run the Service

uvicorn clinicalguard.main:app --host 0.0.0.0 --port 8080

Validate a Clinical Action

curl -X POST http://localhost:8080/ \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tasks/send",
    "params": {
      "id": "task-001",
      "message": {
        "parts": [{
          "text": "validate_clinical_action: Patient SYNTH-042 on Warfarin. Propose Aspirin 325mg daily."
        }]
      }
    }
  }'

Build the App in Python

from clinicalguard.agent import ClinicalGuardApp

guard = ClinicalGuardApp.create(
    constitution_path="constitution/healthcare_v1.yaml",
    audit_log_path="audit.jsonl",
)
app = guard.build_starlette_app()

Key Features

• Healthcare-focused constitutional validation with MACI-style self-approval blocking.
• A2A skills for clinical validation, HIPAA checks, and audit-trail queries.
• Starlette service surface with agent-card, health, and JSON-RPC endpoints.
• Audit-log persistence and restoration for forensic review.
• Optional Anthropic and OpenAI reasoning layers on top of deterministic checks.

Configuration

License

AGPL-3.0-or-later. Commercial licensing is available; contact hello@acgs.ai.

Links

• Homepage
• Documentation
• PyPI
• Repository
• Issues
• Changelog

Constitutional Hash: 608508a9bd224290