Aegis is a state-of-the-art, enterprise-grade AI agent orchestration platform designed to run powerful LLM workflows while preserving absolute data confidentiality. By combining local-first PII anonymization, dynamic client-side re-identification, and cryptographic logging via the Midnight Blockchain, Aegis guarantees strict compliance (HIPAA, GDPR, CCPA) without degrading AI capabilities.

• Dynamic AI Agent Generator: Describe your agent's goal in natural language, and Aegis will automatically generate a fully connected, custom multi-node orchestration pipeline.
• Local Presidio-Grade PII Stripping: Identifies and redacts sensitive patient, financial, or personal information (names, emails, dates, vitals) locally before any data leaves your server.
• Verbatim Client-Side Re-Identification: Re-injects the original patient credentials securely inside your browser, so clinical staff see complete, accurate reports while LLMs remain 100% blind to identities.
• 28 Pre-Integrated Functional Tools: A rich suite of processors spanning Input Nodes (File Upload, Trigger), Transform Utilities (CSV/JSON/Text Splitting), Advanced AI (LLMs, Summarizers, Sentiment), and local code sandboxing.
• On-Chain Cryptographic Auditing: Anchors compliance logs and execution commitments permanently onto the Midnight Blockchain using zero-knowledge proofs. Features live testnet explorer linking.
• Real-Time Telemetry Dashboard: A high-contrast, premium, neo-noir inspired terminal interface that tracks pipeline executions node-by-node.

• Backend: FastAPI, SQLModel, SQLite, high-fidelity local entity anonymization engine
• Frontend: Next.js 14, React, TypeScript, HSL tailored CSS variables, telemetry micro-animations
• Blockchain Integration: Midnight Testnet ZK-proof contract commitments

├── backend/
│   ├── models/           # SQLModel database schemas (Agent, AgentRun, AuditLog)
│   ├── routers/          # FastAPI API endpoint routers (Agents, Audit, LLM)
│   ├── services/         # Core orchestration engines (Workflow Executor, Privacy, LLM Router)
│   └── tools/            # 28 fully functional pipeline processors (llm_prompt, pii_stripper, etc.)
├── frontend/
│   ├── app/              # Next.js App Router (Dashboard, Agent Canvas, Execution Telemetry)
│   ├── components/       # Premium UI components (Terminal, CanvasEditor, WorkflowBuilder)
│   └── lib/              # Keystore vault and API integrations
├── midnight/             # ZK-proof AuditLog smart contracts (AuditLog.compact)
└── docker-compose.yml    # Single-command container deployment

• Python 3.10+
• Node.js 18+

# Initialize Python virtual environment
python -m venv venv
venv\Scripts\activate

# Install dependencies
pip install -r backend/requirements.txt

# Start backend server
uvicorn backend.main:app --reload --port 8000

cd frontend
npm install
npm run dev

Open http://localhost:3000 to view the Aegis dashboard!

graph TD
    A[Raw Scribe Upload] --> B[Local Host: PII Stripper]
    B -->|Anonymized Token Mapping| C[Public Cloud: LLM Pipeline]
    C -->|Secure Context Preservation| D[Structured Report Output]
    D --> E[Local Client: Local Re-Identifier]
    E --> F[Full Verbatim Structured Notes]
    B -->|Compliance State Commitment| G[Midnight Blockchain ZK Log]

Aegis is integrated directly with the Midnight Network to provide unalterable compliance audits:

1. When a workflow run finishes, the local orchestrator constructs an audit state commitment.
2. The state hash is queued and submitted to the Midnight Ledger.
3. Once confirmed, the transaction hash is mapped in your Agent Telemetry panel, providing a verified link to the blockchain explorer.