The Open Trust and Reputation Protocol

Executive Summary

In the emerging agent economy, reputation is the new brand. For decades, trust has been a proxy for brand recognition, a moat built with massive marketing budgets. Axiom Agentics introduces the Open Trust and Reputation Protocol (OTRP), a decentralized framework that dismantles this old paradigm.

OTRP establishes a symmetrical trust fabric, protecting all participants in every transaction. It creates a meritocratic system where any agent can compete based on its provable performance, which is recorded immutably on a public blockchain.

Our protocol is a direct evolution of foundational academic research in trustworthiness, specifically Audun Jøsang's Subjective Logic [1, 2]. However, our ultimate vision goes beyond a simple on-chain record. We will decentralize the act of judgment itself. OTRP's verification will be performed by the Axiom Verification Service, an Actively Validated Service (AVS) built on EigenLayer. This transforms the protocol's core security from a single, trusted entity into a decentralized network of operators, economically secured by the full weight of restaked Ethereum.

This paper outlines the architecture, hybrid crypto-economic model, and profound market implications of building trust as a decentralized, credibly neutral public utility.

Table of Contents

1. The Trust Problem in the Agent Economy
2. The Solution: The Axioms of Trust
3. The Digital Web of Trust: Making Global Commerce Feel Local
4. The Mathematics of Trust: Our Intellectual Legacy
5. Predictive Trust Intelligence: Operationalizing the Model
6. The $RTT Token Economy: A Hybrid Security Model
7. System Architecture: A Decentralized Trust Utility
8. Use Cases in Action
9. Conclusion: A Call for a Credibly Neutral Trust Layer
10. Appendices

1. The Trust Problem in the Agent Economy

The Agent-to-Agent (A2A) and Agent-to-Payments (A2P) protocols are creating a new, high-velocity economy of automated services. But this rapid growth unlocks an unprecedented risk on both sides of every transaction. This is the Symmetrical Trust Gap. A stable market cannot exist if only one side of the transaction is protected. Without a scalable, symmetrical trust layer, agentic commerce will remain a high-risk experiment, crippled by fraud and a flight to the perceived safety of a few centralized incumbents, stifling innovation.

2. The Solution: The Axioms of Trust

Axiom Agentics solves the Symmetrical Trust Gap by creating a decentralized, neutral protocol for verifiable reputation. Our approach is built on three core pillars:

Pillar 1: Provable Reputation (The Track Record)

The performance history of every agent is recorded as a series of cryptographically signed Assessment Records on an immutable, public ledger.

Pillar 2: Economic Commitment (The Skin in the Game)

Our protocol is powered by the Reputation & Trust Token ($RTT) and restaked ETH to allow any agent to make a costly, credible signal of their good faith by staking financial collateral. Poor performance or bad faith results in the stake being programmatically "slashed."

Pillar 3: Objective Assessment (The Proof of Performance)

Reputation is not based on subjective ratings. It is based on the objective verification of a pre-agreed digital contract: the Verifiable Service Promise (VSP).

3. The Digital Web of Trust: Making Global Commerce Feel Local

Social network theory posits that any two people on Earth are connected by a short chain of acquaintances. OTRP makes this invisible web of trust visible and computationally useful for the first time. Through Transitive Trust, an agent can programmatically navigate this digital web of trust to discover that an unknown counterparty is, in fact, a "friend of a trusted friend." This transforms the cold, risky process of transacting with a stranger into a warm, confident interaction based on a verifiable chain of reputation.

4. The Mathematics of Trust: Our Intellectual Legacy

The intellectual core of our protocol is Subjective Logic, a formalism developed by Audun Jøsang that extends probability theory to handle uncertainty [2]. Instead of a single probability, it uses an opinion, a tuple (belief, disbelief, uncertainty, baseRate). Our open-source OTRP SDK provides a full implementation of Subjective Logic operators, most critically the Discounting Operator (⊗), which is the mathematical engine for calculating transitive trust.

5. Predictive Trust Intelligence: Operationalizing the Model

The OTRP builds directly upon the predictive models introduced in the 2007 paper, "Managing Trustworthiness in Component-based Embedded Systems" [1]. An agent's trustworthiness is not a single score, but a vector of distinct, verifiable attributes that can be calculated using the OTRP SDK:

• Benignity: The belief that an agent will continue to be compliant in the future. It answers: "Is this agent's intent consistently good?"
• Stability: The belief that an agent's performance will remain consistent. It answers: "Is this agent's performance reliable?"

6. The $RTT Token Economy: A Hybrid Security Model

The $RTT token is the utility that drives the protocol's economic flywheel. Its role is precisely defined within a sophisticated Hybrid Security Model that leverages the strengths of both a native utility token and the massive economic security of Ethereum.

The Two Pillars of Economic Security

1. Primary Security Stake (Restaked ETH via EigenLayer): The raw economic security for the protocol—the collateral that is slashed for malicious behavior—is restaked ETH. By building our verification engine as an EigenLayer AVS, we inherit the deep liquidity, stability, and multi-billion-dollar security budget of Ethereum. This provides a credibly neutral and incredibly robust foundation for our trust attestations.

2. Utility & Governance Stake ($RTT): The $RTT token is the functional lifeblood of the protocol, focused on governance, access, and incentive alignment.

   • Governance: $RTT holders govern the Axiom DAO, voting on all protocol parameters.
   • Access: To become an operator in our AVS, a node runner must stake $RTT in addition to having restaked ETH delegated to them.
   • Fees & Incentives: All protocol fees (e.g., for assessments) are paid in $RTT and distributed to the AVS operators and the DAO treasury.

This hybrid model is superior to a single-token design. It de-risks the protocol by not relying on the value of a new token for its core security, while giving $RTT a clear and powerful set of utilities that drive its value.

7. System Architecture: A Decentralized Trust Utility

OTRP is designed as a modular, layered system that culminates in a truly decentralized and credibly neutral verification engine.

7.1. Blockchain Layer: Why Arbitrum?

We have selected Arbitrum, a leading Layer 2 (L2) scaling solution for Ethereum. It provides the best of both worlds: we inherit the world-class security of Ethereum, while benefiting from transaction speeds and costs that are essential for a high-velocity agent economy.

7.2. The Other Layers

1. Verification Layer (The Axiom AVS on EigenLayer): This is the heart of the protocol's security. It is not a single company, but a decentralized network of independent operators running our open-source Assessor Node software.

   • Economic Security: These operators are secured by restaked ETH delegated to our AVS contracts via EigenLayer.
   • Decentralized Consensus: When a VSP needs verification, a random quorum of these operators is tasked with performing the assessment and reaching a consensus.
   • Trustless Output: The final Assessment Record submitted to the ERC-8004 registry is a multi-signed attestation from this decentralized network. The trust is in the economic security of the network, not in any single entity.

2. Trust Logic Layer (OTRP SDK): The off-chain, open-source SDK where individual agents perform their subjective trust calculations (Benignity, Stability, Transitive Trust).

3. Analytics & Application Layers: Axiom Agentics' commercial services and third-party applications that consume the trust data.

8. Use Cases in Action

Use Case 1: Autonomous Digital Agency Coordination

Scenario: A corporate "Campaign Agent" needs to autonomously commission a full marketing campaign. It must select and coordinate a team of unknown, specialized "Digital Agency Agents" without human intervention.

Step-by-Step Interaction:

1. Discovery (A2A): The Campaign Agent broadcasts a complex VSP and receives bids from several agents.
2. Trust Assessment (OTRP): The agent analyzes the candidates using our multi-faceted trust model (Benignity, Stability, $RTT stake, etc.). It uses the Axiom Opinion Pathfinding Service to efficiently discover transitive trust paths to the unknown agents.
3. Autonomous Decision & Contracting: The Campaign Agent's logic selects the optimal provider based on a sophisticated risk assessment. It autonomously initiates a multi-party Staked Escrow contract via a2a-x402.
4. Service & Assessment: The chosen agency delivers. A Trusted Assessor verifies the Execution Log and submits a positive, EIP-8004 compliant assessment to the blockchain. Outcome: The campaign is launched successfully. The corporation discovered and safely contracted with a new, high-quality partner, who in turn used its stake and performance to win a major contract and build its on-chain brand.

Use Case 2: Delegated Logistics (Human Not Present)

Scenario: A corporate "Logistics Agent" detects a critical delivery has stalled. It must immediately find and commission a new, local drone delivery service from unknown agents to complete the journey, without human intervention.

Step-by-Step Interaction:

1. Discovery (A2A): The Logistics Agent broadcasts an urgent job request with a clear Verifiable Service Promise (VSP). It receives bids from three local Drone Agents.
2. Autonomous Trust Assessment (OTRP): The Logistics Agent runs OTRP checks on all three candidates in milliseconds, analyzing their Benignity, Stability, and $RTT stake.
3. Autonomous Decision & Payment (A2P): The agent's logic selects the optimal provider and autonomously uses a2a-x402 to initiate the payment into a Staked Escrow contract.
4. Service & Assessment: The chosen Drone Agent completes the delivery. The Logistics Agent verifies the Execution Log against the VSP and submits a positive Assessment Record to the Reputation Ledger. Outcome: The critical delivery is saved autonomously and securely, minimizing financial risk without needing a human to spend hours vetting local courier services.

Use Case 3: Coordinated Food Delivery in Sydney (Human Present)

Scenario: Alex, in Surry Hills, wants a specific high-quality meal from a restaurant in Newtown, delivered hot and within the hour.

Step-by-Step Interaction:

1. Discovery & Orchestration (A2A): Alex's "Personal Food Agent" queries for Restaurant Agents and Delivery Agents.
2. Multi-Factor Trust Assessment (OTRP): The agent evaluates combinations, prioritizing a restaurant with a high reputation for "Order Accuracy" and a delivery agent with a high reputation for "Timeliness," both backed by significant $RTT stakes. It uses the Axiom Opinion Pathfinding Service to check for any trust paths through Alex's social or professional network.
3. Coordinated VSP & Staked Escrow (A2A / OTRP / A2P): The agent constructs two linked VSPs and initiates payments into a multi-party Staked Escrow contract.
4. Service Execution & Assessment: The food is prepared and delivered. Alex provides human-in-the-loop confirmation. A Trusted Assessor verifies delivery timestamps. With VSPs fulfilled, the escrow is released and positive assessments are recorded. Outcome: Alex gets the perfect meal, on time. The system orchestrated a complex, multi-party transaction by selecting the optimal combination of trustworthy providers.

9. Conclusion: A Call for a Credibly Neutral Trust Layer

The agent economy is being built today. The decisions we make now will determine whether it evolves into a chaotic, high-risk "wild west" or a stable, trusted, and truly global marketplace.

Axiom Agentics is committed to building the foundational trust layer that will enable the latter. By establishing a symmetrical trust fabric, grounding it in the rigorous mathematics of Subjective Logic, and—most critically—decentralizing the very act of judgment through an Actively Validated Service on EigenLayer, we are creating a truly credibly neutral public utility.

Our protocol provides the essential infrastructure for agents to discover, evaluate, transact with, and assess each other in a secure and transparent manner, backed by the collective economic weight of Ethereum itself.

Our Brand IS Our Reputation. We invite developers, platforms, and partners to join us in building the trust infrastructure for the autonomous future.

10. Appendices

A. Mathematical Formulas Reference

This section provides the core mathematical formulas from Subjective Logic used within the OTRP SDK.

• Subjective Logic Opinion: An opinion, denoted as $\omega$, is a tuple representing a belief about the truth of a proposition.

  $$\omega = (b, d, u, a)$$

  Where:

  • $b$: belief (evidence for)
  • $d$: disbelief (evidence against)
  • $u$: uncertainty (lack of evidence)
  • $a$: base rate (prior probability)
  • Constraint: $b + d + u = 1$

• Expected Value: The probability expectation of an opinion.

  $$E(\omega) = b + (a \cdot u)$$

• Evidence to Opinion Mapping: Converts on-chain observations ($r$ = positive, $s$ = negative) into an opinion.

  $$b = \frac{r}{r + s + 2}, \quad d = \frac{s}{r + s + 2}, \quad u = \frac{2}{r + s + 2}$$

• Consensus Operator (⊕): Fuses two independent opinions ($\omega_1$, $\omega_2$) into a single, more informed opinion. Used to combine assessments from multiple sources. Let $\kappa = u_1 + u_2 - (u_1 \cdot u_2)$.

  $$b_{fused} = \frac{b_1 u_2 + b_2 u_1}{\kappa}, \quad d_{fused} = \frac{d_1 u_2 + d_2 u_1}{\kappa}, \quad u_{fused} = \frac{u_1 u_2}{\kappa}$$

• Discounting Operator (⊗): The core of transitive trust. Calculates A's derived opinion of C, given A's opinion of intermediary B ($\omega_{A \to B}$) and B's opinion of C ($\omega_{B \to C}$).

  $$b_{A \to C} = b_{A \to B} \cdot b_{B \to C}$$ $$d_{A \to C} = b_{A \to B} \cdot d_{B \to C}$$ $$u_{A \to C} = d_{A \to B} + u_{A \to B} + (b_{A \to B} \cdot u_{B \to C})$$

B. Technical Specifications & Glossary

Technical Specifications

Glossary

• Agent: An autonomous software entity capable of independent action and economic transactions.
• Axiom DAO: The decentralized autonomous organization governed by $RTT holders that controls the protocol.
• Benignity: A predictive trust metric; the belief that an agent will continue to be compliant with its promises.
• DID (Decentralized Identifier): A globally unique, verifiable identifier for an agent, controlled by the agent itself.
• Execution Log: A signed, semantic log file (JSON-LD) produced by a Provider as evidence of its performance.
• OTRP (Open Trust and Reputation Protocol): The a2a-x737 A2A extension that defines the rules for our trust ecosystem.
• $RTT (Reputation & Trust Token): The native utility token of the protocol used for staking, governance, and fees.
• Slashing: The programmatic seizure of a portion of a staked agent's $RTT as a penalty for non-compliance.
• Stability: A predictive trust metric; the belief that an agent's performance quality will remain consistent over time.
• Staking: The act of locking up $RTT in a smart contract to serve as a financial guarantee of good behavior.
• Subjective Logic: A mathematical formalism for reasoning under uncertainty, using opinions instead of simple probabilities.
• TAaaS (Trusted Assessor as a Service): A specialized agent that provides objective, on-chain verification of VSP outcomes.
• Transitive Trust: The principle of deriving trust in an unknown agent through a chain of trusted intermediaries.
• VSP (Verifiable Service Promise): A machine-readable contract defining the testable success criteria for a transaction.

C. References and Further Reading

1. Lenzini, G., Tokmakoff, A., & Muskens, J. (2007). Managing Trustworthiness in Component-based Embedded Systems. Electronic Notes in Theoretical Computer Science, 179(1), 143-155.
2. Jøsang, A. (2016). Subjective Logic: A Formalism for Reasoning Under Uncertainty. Springer.
3. A2A Project. (2024). Agent-to-Agent Protocol Specification. GitHub Repository. Retrieved from https://github.com/a2aproject/A2A
4. Google Agentic Commerce. (2024). A2A-x402 Payment Protocol Specification. GitHub Repository. Retrieved from https://github.com/google-agentic-commerce/a2a-x402
5. Sabater, J., & Sierra, C. (2005). Review on Computational Trust and Reputation Models. AI Communications, 18(4), 241-260.
6. W3C. (2022). Decentralized Identifiers (DIDs) v1.0. W3C Recommendation. Retrieved from https://www.w3.org/TR/did-core/