Skip to content

rumagoso/TRACE

Repository files navigation

TRACE — Trusted Registry for Autonomous and Connected Entities

A proposed open infrastructure standard for AI agent identity and accountability.


The Problem

AI agents book meetings, execute financial transactions, control industrial systems, and spawn sub-agents — across organisational and jurisdictional boundaries — with no shared infrastructure to answer a basic question:

Who are you, who sent you, and what are you authorised to do?

Enterprise IAM standards (OAuth 2.0, SPIFFE/SPIRE, OpenID Connect) address agents operating inside a single organisation's trust perimeter. They were not designed for agents crossing that perimeter. No current standard was.

This is not a future risk. It is a present condition.


What TRACE Is

TRACE is a proposed four-layer open standard to close this gap.

Layer Name Function
1 Identity Cryptographic identity bound to a verified legal entity, recorded on a permissioned ledger using W3C DIDs. Recursive Sub-Tokens enable high-velocity delegation without losing chain of custody.
2 Access Resources query TRACE before granting access. Registered and in good standing: access at the appropriate trust tier. Unregistered: no access, without exception.
3 Governance The TRACE Foundation — a multi-stakeholder non-profit modelled on ICANN — governs the registry at two speeds: automated Circuit Breakers for real-time containment, human supermajority for permanent revocation.
4 Economic Alignment High-risk agents post conditional collateral held in escrow. On verified failure, collateral is released to a victim recourse pool. Clean records earn tiered infrastructure incentives.

TRACE complements, not competes with, existing frameworks. It provides the identity substrate the EU AI Act assumes but does not specify; the accountability chain ISO 42001 requires but does not provision; the revocation mechanism NIS2 implies but does not mandate for AI.


Status

Milestone Status
Position Paper v0.2 Published — March 2026
NIST NCCoE Public Comment Submitted — March 2026
CSA Blog Submission Pending
Working Group Formation Open for interest

Documents

  • Position Paper v0.2LinkedIn — full architecture, threat model, standards alignment, and call to action
  • NIST NCCoE Public Comment — submitted to AI-Identity@nist.gov ahead of the April 2, 2026 deadline — available in this repository

Standards Alignment

TRACE is designed for compatibility with: W3C DID Core 1.0, OAuth 2.0, NIST SP 800-63-4, NIST AI RMF, ISO/IEC 42001, EU AI Act (Articles 9, 13, 22), NIS2 Directive, EBSI (EU deployments), CSA AICM.


Contributing

See CONTRIBUTING.md.


Author

Rui Soares ISMS Manager, Crossjoin Solutions | Invited Lecturer, NOVA IMS, Lisbon | CISSP rui.msoares@gmail.com


Licence

Creative Commons Attribution 4.0 International (CC BY 4.0)

You are free to share and adapt this material for any purpose, including commercial, provided you give appropriate credit.

About

TRACE - Trusted Registry for Autonomous and Connected Entities — For autonomous AI traceability. An open infrastructure standard specifically designed to address the cross-boundary identity and accountability gap.

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors