██████╗ ██████╗ ██████╗ ████████╗██╗
 ██╔════╝██╔═══██╗██╔══██╗╚══██╔══╝██║
 ██║     ██║   ██║██████╔╝   ██║   ██║
 ██║     ██║   ██║██╔══██╗   ██║   ██║
 ╚██████╗╚██████╔╝██║  ██║   ██║   ██║
  ╚═════╝ ╚═════╝ ╚═╝  ╚═╝   ╚═╝   ╚═╝

  █████╗  ██████╗ ███████╗███╗   ██╗████████╗███████╗
 ██╔══██╗██╔════╝ ██╔════╝████╗  ██║╚══██╔══╝██╔════╝
 ███████║██║  ███╗█████╗  ██╔██╗ ██║   ██║   ███████╗
 ██╔══██║██║   ██║██╔══╝  ██║╚██╗██║   ██║   ╚════██║
 ██║  ██║╚██████╔╝███████╗██║ ╚████║   ██║   ███████║
 ╚═╝  ╚═╝ ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ╚═╝   ╚══════╝

A community-driven library of production-ready AI agent system prompts for healthcare and life sciences, built on the Corti Agentic Framework.

📖 Table of Contents

• About This Repository
• Key Features
• Repository Structure
• Quick Start
• Contributing
• Security and Privacy

🏥 About This Repository

This repository serves as a prompt library and validation toolkit for AI agents in healthcare. It provides:

1. Curated System Prompts: Production-ready, field-tested agent configurations for clinical and administrative workflows
2. Standardized Schema: A structured YAML format ensuring consistency and maintainability
3. Validation Tools: Automated quality checks for prompt formatting, readability, and best practices
4. Community Contributions: An open platform for healthcare organizations and developers to share and improve agent prompts

What is the Corti Agentic Framework?

The Corti Agentic Framework is a specialized AI platform designed specifically for healthcare and life sciences, focused on creating reliable, governed multi-agent systems. It transforms "agent chaos into governed systems" by providing:

• Governed Orchestration: Manages agent roles, routes tasks, and enforces boundaries
• Domain-Specific Expertise: Pre-configured "Experts" with targeted clinical and administrative intelligence
• Comprehensive Oversight: Guardrails, audit trails, and complete visibility into agent actions
• Standards Support: Built on Agent-to-Agent (A2A) and Model Context Protocol (MCP) standards
• High-Stakes Reliability: Designed for healthcare environments requiring predictable and compliant AI behavior

This repository contains the system prompts that power these intelligent agents, enabling capabilities like medical coding, clinical documentation improvement (CDI), revenue cycle management, and automated clinical workflows.

🎯 Key Features

For Prompt Library Users

• 🤖 Production-Ready Agents: Field-tested prompts for medical coding, clinical intelligence, and administrative tasks
• 📋 Standardized Format: Consistent YAML schema across all agent configurations
• 📚 Comprehensive Documentation: Detailed descriptions, use cases, and execution flows for each agent
• ✅ Quality Assurance: All prompts validated for syntax, structure, and readability

For Contributors

• 🛠️ Validation Toolkit: Automated YAML validation with comprehensive checks
• 📖 Clear Schema: Well-documented structure for creating new agent prompts
• 🎨 Best Practices: Built-in checks for readability, formatting, and quotation standards
• 🔄 Version Control: Semantic versioning for prompt iterations
• 🤝 Open Collaboration: Community-driven improvement and expansion

📂 Repository Structure

corti-agents/
├── prompts/                     # Agent system prompt library
│   ├── schema.yaml              # YAML schema specification
│   └── [additional agents]      # Community-contributed prompts
├── tools/                       # Validation toolkit
│   └── validate_prompts.py      # Comprehensive YAML validator
└── README.md                    # This file

🚀 Quick Start

For Prompt Users

1. Browse Available Agents: Explore the prompts/ directory for production-ready agent configurations
2. Select an Agent: Choose an agent that matches your use case (e.g., medical coding, clinical documentation)
3. Integrate with Framework: Use these prompts with the Corti Agentic Framework by integrating directly with the Corti API (docs.corti.ai) or on the Console (console.corti.app)
4. Customize as Needed: Adapt prompts to your specific requirements while maintaining the schema structure

For Contributors

Prerequisites

• Python 3.7 or higher
• PyYAML library

pip install pyyaml

Creating a New Agent Prompt

1. Use the Schema: Follow the structure defined in prompts/schema.yaml
2. Reference Examples: Use medical-coding.yaml as a template
3. Required Fields: Ensure all mandatory schema fields are included
4. Best Practices: Use literal block scalars (|) for long text, avoid unnecessary quotes
5. Validate: Run the validation tool before submitting

Validating Your Prompts

Validate all prompt files against the schema:

python tools/validate_prompts.py

The validator checks for:

• ✅ YAML syntax errors
• ✅ Schema compliance (required fields)
• ✅ Extra fields not defined in schema
• ✅ Proper string quoting and escaping
• ✅ Mismatched or unclosed quotations
• ✅ Indentation consistency
• ✅ Special character handling
• ✅ Field type validation
• ✅ Quotation best practices (warns about unnecessary quotes)
• ✅ Readability best practices (long strings, literal blocks, paragraph breaks)

🤝 Contributing to the Library

We welcome contributions from healthcare organizations, AI developers, and clinical informaticists! This library thrives on community input and real-world experience.

Contribution Guidelines

• Production-Ready: Only submit agents tested in real workflows
• Schema Compliance: All agents must pass validation
• Clear Documentation: Provide detailed descriptions and use cases
• Semantic Versioning: Use major.minor.patch versioning (e.g., 1.0.0)
• Expert Attribution: Credit your organization in the builder field
• Privacy First: Never include PHI, credentials, or proprietary information

🔐 Security and Privacy

This repository contains system prompts only - no patient data, credentials, or proprietary information should ever be included. When contributing:

• ✅ Share agent instructions and workflow logic
• ✅ Describe use cases and expected inputs
• ✅ Document output formats and structures
• ❌ Never include PHI or PII
• ❌ Never include API keys or credentials
• ❌ Never include institution-specific data

📄 License

Apache 2.0

🌟 Acknowledgments

Built by the healthcare AI community for advancing reliable, governed agent systems in clinical and administrative workflows.