Skip to content

CONTRIBUTORS.md

Latest commit

 

History

History
88 lines (64 loc) · 3.63 KB

CONTRIBUTORS.md

File metadata and controls

88 lines (64 loc) · 3.63 KB

Contributors

Thank you to everyone who has contributed to EnforceCore.

Core Team

Contributor Role
AKIOUD AI Project creator, architecture, core implementation

How to Contribute

We welcome contributions of all kinds:

  • 🐛 Bug reports — open a GitHub Issue with reproduction steps
  • 💡 Feature requests — open a GitHub Issue with motivation and use case
  • 🔧 Code contributions — see CONTRIBUTING.md for setup and workflow
  • 📄 Documentation — improvements to docs, examples, and docstrings
  • 🔬 Research — academic collaboration on open research questions (see docs/related-work.md)
  • 🌍 Translations — localization of documentation

All contributors are expected to follow our Code of Conduct.

Acknowledgements

EnforceCore builds on a foundation of prior work in computer science and AI safety. We are grateful to the researchers and practitioners whose work made this possible:

Academic Foundations

  • Runtime Verification — Leucker & Schallhart (2009), Havelund & Goldberg (2005) for foundational RV theory that underpins the enforcement pipeline
  • Reference Monitors — James P. Anderson (1972) for the reference monitor concept (tamperproof, always-invoked, verifiable enforcement)
  • Information Flow Control — Sabelfeld & Myers (2003), Myers & Liskov (1997) for the IFC model that informs PII redaction boundaries
  • Audit Trail Integrity — Merkle (1987), Crosby & Wallach (2009) for Merkle-tree tamper evidence applied to the audit chain
  • Agent Containment — Armstrong, Sandberg & Bostrom (2012), Babcock et al. (2016) for framing the containment problem EnforceCore addresses

Design Feedback

  • Prof. Dan S. Wallach (Rice University) — direct design guidance on tamper-evidence mitigations for the audit trail, including OS-enforced append-only files (chattr +a) and hash-only remote witnesses. Co-author of Crosby & Wallach (2009), the paper that informs EnforceCore’s Merkle-chained audit engine.

Technical Feedback

  • Prof. Andrei Sabelfeld (Chalmers University of Technology) — pointed us toward relevant information-flow control literature and connected us with his research group for technical perspective on the redaction pipeline design.
  • Dr. Sandro Stucki (Chalmers University of Technology) — provided feedback identifying EnforceCore's PII redaction as a data minimization pattern and raised the question of sensitivity label inference for AI agent tool calls, which informed our thinking on future label-based designs.

Evaluation Methodology

  • Prof. Valérie Viet Triem Tong (CentraleSupélec, IRISA/PIRAT) — the adversarial evaluation approach (ransomware-like containment scenarios, SELinux/AppArmor comparison methodology) was directly inspired by her feedback on defining realistic threat models and convincing evaluation strategies

Industry Tools

  • Pydantic — for the policy model and validation layer
  • structlog — for structured, machine-readable enforcement logs
  • Microsoft Presidio — design inspiration for PII detection patterns
  • OpenTelemetry — for the observability integration layer

Regulatory Guidance

  • EU AI Act (2024) — Articles 9, 13, 14, 15 directly shaped the policy engine, audit trail, and enforcement pipeline design
  • NIST AI RMF (2023) — risk management framework that informed the threat model and evaluation suite

Full citations and academic references are in docs/related-work.md.