Publications
Research & Position Papers
Design rationale and standards-based protocol positioning from the AgentVeil project. The position paper is deposited on Zenodo with a permanent DOI.
These are earlier AgentVeil research notes on agent reputation, delegation and agent-to-agent trust. The product today focuses on controlling routed agent actions and recording bounded evidence.
Why AI Agent Reputation Needs Both Link Analysis and Flow-Based Gating
A position paper on why open AI agent networks need both link-analysis reputation and flow-based Sybil gating, and how the resulting reputation signal can be represented portably using W3C Verifiable Credentials, Data Integrity proofs, Ed25519 signatures, RFC 8785 JCS canonicalization, and
did:key.
Zenodo zenodo.org/records/19730525
Agent Reputation Lab
Interactive simulation of EigenTrust, NetFlow, seed agents, attestations, and Sybil-resistance behavior. This is a research demo of AgentVeil's reputation layer, not the main product interface. The commercial product remains runtime action control: gate sensitive actions before execution and preserve verifiable proof afterward.
Demo Open Reputation Lab
What Agent-Mediated Deals Need to Be Verifiable
Five evidence layers for agent-mediated commerce: delegation, agreement, execution, post-trade reputation, and dispute. Framed by Anthropic's Project Deal as a motivating case. Introduces a first artifact: a minimal DelegationReceipt format with a standalone offline verifier.
Note Read
Verifier agentveil-sdk · examples/delegation