The Aegis Protocol is a modular, AI-driven backend for auditing and defending healthcare claims. It operates with a "Neuro-Core" using Google's Gemini to classify and resolve claim issues, enforcing strict Pydantic models for all data flow.

• Agents:
  • Classifier (Neuro): Analyzes claims using Gemini to predict rejection/denial codes.
  • Resolver (Neuro): Auto-fixes JSON errors or drafts defense letters for medical necessity denials.
  • Submitter: Simulates API submission.
  • Watchdog: Monitors payer responses and handles recursion.
• Core:
  • Registry: Interface to the local code database (TinyDB).
  • Models: Pydantic definitions for Claims, Reports, etc.
  • LLM: Abstracted Gemini client (JSON Mode).

1. Install Dependencies:

   pip install -r requirements.txt

2. Environment Variables: Create a .env file in the root aegis_core/ directory:

   GOOGLE_API_KEY=your_gemini_key_here

3. Seed Database (Required): Hydrate the local "Source of Truth" database:

   python scripts/seed_registry.py

4. Run the Protocol:

   python -m main

• agents/: Autonomous agent logic.
• core/: Shared utilities and data models.
• data/: Local database storage (ignored by git).
• scripts/: Setup scripts.
• main.py: Entry point.