ContextGuard-MCP-PQC

Post-Quantum Security Framework for Model Context Protocol Using ContextGuard

Implementation artifacts for the ICSCSA 2026 paper presenting the first systematic application of RFC 9794 Post-Quantum/Traditional hybrid terminology to MCP's four security-relevant layers, and the ContextGuard-MCP-PQC governance framework.

Paper

Title: Post-Quantum Security Framework for Model Context Protocol Using ContextGuard

Authors:

1. Sunil Gentyala -- Independent Researcher, HCLTech America Inc., Dallas TX, USA (IEEE Senior Member #101760715)
2. P. Krishna Sunil -- Assistant Professor, Dept. of MBA, Narasaraopeta Engineering College, Narasaraopet, India
3. Vamshi Lande -- Research Student, University of North Texas, Denton TX, USA
4. Akhila Kasturi -- Research Analyst Lead, HCLTech, Dallas TX, USA
5. Suresh Kumar Darisi -- Senior Systems Engineer, Rocket Software Inc., Boston, USA
6. Ruhisulthana Shaik -- Lecturer, Dept. of ECE, Sir C.R. Reddy Polytechnic College, Eluru, India

Conference: 2026 6th International Conference on Soft Computing for Security Applications (ICSCSA), Aug 19-21, 2026. Accepted, paper ICSCSA-011; IEEE Xplore DOI pending.

Repository Contents

contextguard-mcp-pqc/
├── poc/
│   ├── hndl_demo.py          Harvest-Now-Decrypt-Later threat simulation
│   └── downgrade_demo.py     TLS cipher suite downgrade attack demo
├── benchmarks/
│   ├── pqc_bench.py          ML-KEM / ML-DSA performance benchmarks
│   └── results/              Benchmark output (CSV + JSON)
├── risk/
│   └── risk_assessment.py    Quantitative vulnerability risk scoring
├── framework/
│   ├── haf.py                Hybrid Authentication Floor (Layer 1)
│   └── hrs.py                Hybrid Readiness Score calculator (Layer 3)
├── CITATION.cff
├── LICENSE
└── README.md

Framework Architecture

ContextGuard-MCP-PQC extends the base ContextGuard zero-trust governance architecture with a post-quantum control plane aligned to RFC 9794 constructs.

Vulnerability Classes

Quick Start

git clone https://github.com/sunilgentyala/contextguard-mcp-pqc
cd contextguard-mcp-pqc
pip install -r requirements.txt

# Run HNDL threat demonstration
python poc/hndl_demo.py

# Run downgrade attack demonstration
python poc/downgrade_demo.py

# Run performance benchmarks
python benchmarks/pqc_bench.py

# Run risk assessment
python risk/risk_assessment.py

# Run HRS calculator demo
python framework/hrs.py

Optional: For live ML-KEM / ML-DSA benchmarks, install liboqs-python:

pip install oqs-python

Performance Summary

Reference values from Olushola and Meenakshi (Frontiers in Physics, 2026,
DOI: 10.3389/fphy.2025.1723966) and Halak et al. (MDPI Cryptography, 2024,
DOI: 10.3390/cryptography8020021):

Composite hybrid TLS handshake adds approximately 2,272 bytes of overhead per
connection (Stebila et al., draft-ietf-tls-hybrid-design-12, 2025).

HRS Certificate Scoring Criteria

Score 0 = Traditional-only (Priority 1, immediate action). Score 5 = Fully hybrid.

Contributing

Issues and pull requests are welcome — new vulnerability classes, live benchmark data, a Layer 4 (RSP) implementation, and a reference MCP server that actually enforces HAF/HKF are all open areas. See CONTRIBUTING.md for setup and PR guidelines.

Citation

@inproceedings{gentyala2026contextguard_pqc,
  title     = {Post-Quantum Security Framework for Model Context Protocol Using ContextGuard},
  author    = {Gentyala, Sunil and Krishna Sunil, P. and Lande, Vamshi
               and Kasturi, Akhila and Darisi, Suresh Kumar and Shaik, Ruhisulthana},
  booktitle = {2026 6th International Conference on Soft Computing for
               Security Applications (ICSCSA)},
  publisher = {IEEE},
  year      = {2026},
  note      = {Paper ICSCSA-011. Accepted; IEEE Xplore DOI pending.},
  url       = {https://github.com/sunilgentyala/contextguard-mcp-pqc}
}

License

MIT License. See LICENSE for details.

This repository contains implementation artifacts and proof-of-concept
demonstrations for academic research. The PoC code in poc/ is for
educational purposes only.