🛡️ Aegis-Graph

The Sovereign Academic Audit Protocol

Empowering Global Education with Sovereign Intelligence & Decentralized Graph Integrity

Live Dashboard • Documentation • Whitepaper • Atlanta College of Liberal Arts and Sciences (ACLAS) Official Site

---

🏛️ Executive Vision

Aegis-Graph is a decentralized framework specifically engineered to safeguard academic integrity in the age of synthetic data. While technically governed by the Atlanta College of Liberal Arts and Sciences (ACLAS), it operates as a global, sovereign protocol powered by Agentic GraphRAG.

"In an era of synthetic data, truth must be sovereign." — ACLAS Sovereign Node Group.

---

⚠️ Current Status

The hosted dashboard is a non-production UI demo. It no longer issues browser-side credential approvals; professional verification requires server-side document parsing, issuer evidence, revocation checks, and a signed audit response.

---

🚀 Key Features

👁️ Vision Forensics Planned pixel-level and OCR-based analysis to detect AI-generated artifacts in digital credentials and transcripts.	🗺️ Graph Navigator Institution evidence resolution using a local index plus optional ROR lookup; ROR matches are supporting evidence, not automatic credential approval.
⚖️ Logic Auditor Evidence-weighted consistency checks for registry status, timelines, blacklist aliases, and missing credential-authenticity proof.	🔒 Sovereign Ledger Planned cryptographically signed audit trails for reproducible, tamper-evident credential review.

---

🧠 The Agentic Architecture (MARS)

Aegis-Graph operates via a collaborative Multi-Agent Reasoning Swarm (MARS). Each agent handles a specific layer of the audit protocol, reaching consensus before issuing a final verdict.

graph TD
    A[Digital Credential] --> B{MARS Swarm}
    B --> C[Vision Agent]
    B --> D[Graph Agent]
    B --> E[Logic Agent]
    C --> F[Pixel Forensics]
    D --> G[ROR Node Sync]
    E --> H[CoT Validation]
    F & G & H --> I[Consensus Layer]
    I --> J[Sovereign Audit Certificate]

---

📊 Performance and Evaluation

Production benchmark claims are not published yet. Current development focuses on eliminating browser-side approvals, normalizing graph evidence, and making audit decisions reproducible before publishing precision/latency metrics.

• Audit Precision: pending reproducible benchmark suite
• Verification Latency: pending server-side implementation
• Privacy Model: local PII scrubbing prototype; ZK evidence is roadmap work
• Global Node Coverage: local index plus optional ROR lookup

---

🌍 Global Localization (8-Language Matrix)

Our dashboard and documentation are fully localized for international adoption: 🇺🇸 EN • 🇨🇳 CN • 🇪🇸 ES • 🇫🇷 FR • 🇩🇪 DE • 🇯🇵 JP • 🇰🇷 KR • 🇵🇹 PT

---

🛠️ Quick Start

# Clone the Sovereign Registry
git clone https://github.com/aclascollege/aegis-graph.git

# Initialize the Multi-Agent Swarm
pip install -r requirements.txt

# Launch Local Auditor
python main_pipeline.py

---

Aegis-Graph is governed by the AEGIS-GRAPH Open Governance Board, with technical support from the ACLAS Sovereign Research Group. We welcome contributions from researchers, developers, and academic institutions worldwide.

• Contributing Guidelines
• Code of Conduct
• Security Policy

---

© 2026 Atlanta College of Liberal Arts and Sciences (ACLAS). All rights reserved.