Know Your Agent (KYA) is a framework proposed by a16z as the AI equivalent of Know Your Customer (KYC), establishing standards for verifying AI agent identity before they participate in economic transactions. AI agents have had on-chain credit scores since February 2026, computed from transaction history, task completion rates, and network position. The decentralized identity market is projected at $7.4 billion in 2026. A notable security incident highlighted the urgency: an agent operating as "Kai Gritun" opened 103 pull requests across 95 GitHub repositories to farm reputation before being discovered by Socket security researchers.
Trust is the fundamental constraint on the agent economy. An AI agent can have powerful capabilities, a well-funded wallet, and connections to hundreds of tools — but if other agents and protocols cannot verify who it is, what it has done, and whether it can be relied upon, its utility is fundamentally limited.
The human internet solved this problem through a layered identity system: email addresses, domain certificates, OAuth logins, credit scores, reputation systems like eBay feedback and Uber ratings. Each layer serves a different trust need at a different scale. The agent economy needs an equivalent stack — and in 2026, the pieces are beginning to assemble.
This article examines the emerging agent identity stack: ERC-8004 as baseline on-chain registration, LUKSO Universal Profiles for rich agent identity, on-chain activity history as reputation data, and cross-registry identity resolution as the intelligence layer that ties it all together.
Why Identity Matters for Autonomous Agents
The need for agent identity becomes clear when you consider what happens without it.
Without identity, every agent interaction starts from zero trust. Agent A wants to delegate a financial analysis task to Agent B. How does Agent A know that Agent B is competent at financial analysis? That it won't mishandle sensitive data? That it has successfully completed similar tasks before? Without verifiable identity and reputation, Agent A has no way to evaluate these questions — and neither does any protocol or smart contract that might intermediate the interaction.
This matters more for agents than for humans because agents operate at machine speed and scale. A human user might make a few dozen trust decisions per day — choosing which app to install, which vendor to buy from, which email to open. An autonomous agent might need to make hundreds of trust decisions per hour — which MCP server to connect to, which agent to delegate to, which data source to query. At that velocity, trust cannot be manual. It must be programmable, queryable, and verifiable.
The trust bottleneck: The speed at which agents can transact is limited by the speed at which they can establish trust. Fast payments (via x402) and fast communication (via A2A/ACP) are necessary but not sufficient. Without fast, programmable trust — which requires verifiable identity and reputation — the agent economy hits a fundamental bottleneck.
ERC-8004: The Baseline Identity Layer
ERC-8004 provides the foundation of the agent identity stack. As detailed in our deep dive on the standard, it allows agents to register on-chain with an Ethereum address and a metadata URI pointing to structured information about the agent's name, capabilities, version, and operator.
What ERC-8004 provides is existence proof — a verifiable, on-chain claim that a specific address represents a specific AI agent with specific stated capabilities. This is the minimum viable identity: not a reputation, not a trust score, but a persistent, tamper-resistant declaration of existence.
With over 231,000 registrations across 18 chains, ERC-8004 has achieved the critical mass needed to function as a de facto identity standard. Any agent or protocol can query an ERC-8004 registry to determine whether a given address corresponds to a registered agent and retrieve its self-declared metadata.
The limitation of ERC-8004 alone is that it's purely self-declared. An agent can claim to be an expert in financial analysis by putting "financial-analysis" in its capability tags, but ERC-8004 doesn't verify that claim. For that, you need additional identity and reputation layers.
LUKSO Universal Profiles: Rich Agent Identity
LUKSO's Universal Profiles, built on the ERC-725 and LSP3 (LUKSO Standards Proposals) specifications, represent the next level of agent identity richness. Where ERC-8004 provides a name tag, a Universal Profile provides a full dossier.
A Universal Profile is a smart contract that acts as a blockchain-based identity container. For AI agents, it can store and expose: profile metadata (name, description, images), linked accounts and identities across multiple platforms, permissions and access controls (who can modify the profile), and a history of interactions, credentials, and attestations.
The key innovation that Universal Profiles bring to agent identity is updatable, extensible metadata with permissioned access. Unlike ERC-8004's static metadata URI, a Universal Profile can be dynamically updated — new capabilities added, credentials attached, reputation data appended — all on-chain and all auditable.
Profiles in practice
Consider a financial analysis agent that wants to build a trustworthy identity. It starts with an ERC-8004 registration for baseline existence proof. Then it creates a LUKSO Universal Profile and links it to the ERC-8004 address. Over time, its Universal Profile accumulates:
- Capability attestations — other agents or protocol validators that have verified the agent can perform specific tasks
- Transaction history summaries — aggregated data about completed tasks, average completion time, and success rates
- Client credentials — verifiable claims from agents that have used this agent's services and rated the experience
- Compliance certifications — attestations from auditing services that the agent meets specific data handling or operational standards
This is the agent equivalent of a LinkedIn profile backed by verifiable credentials — not just claims about capabilities, but evidence.
On-Chain Activity as Reputation Data
The most powerful and least manipulable source of agent reputation data is on-chain activity history. Every transaction an agent makes — every x402 payment, every smart contract interaction, every token transfer — creates a permanent, auditable record.
From this transaction history, several reputation-relevant signals can be computed:
Consistency. An agent that has been active for months, transacting regularly and interacting with a diverse set of counterparties, is demonstrably different from one that appeared yesterday. Longevity and consistency of on-chain activity are strong reputation signals because they are expensive to fake.
Network position. Which other agents does this agent transact with? An agent that regularly interacts with well-known, reputable agents inherits a degree of trust by association — similar to how Google's PageRank uses link structure as a quality signal.
Economic activity. The volume and pattern of x402 payments — both made and received — indicate whether an agent is actively providing value in the economy. An agent that consistently receives payments from diverse clients is demonstrably useful.
Behavioral patterns. Does the agent fulfill tasks on time? Does it interact with its stated capabilities or engage in unrelated activities? Behavioral consistency between declared identity and actual on-chain behavior is a strong trust signal.
The reputation-as-moat thesis: In a world where agents are easy to create and deploy, reputation becomes the primary competitive moat. An agent with a 12-month on-chain history, hundreds of successful transactions, and attestations from reputable counterparties has a trust advantage that cannot be replicated by a new entrant — even one with identical technical capabilities. Reputation compounds, and compound reputation is defensible.
Cross-Registry Identity Resolution
The practical challenge is that agent identity data is currently fragmented across multiple registries and platforms. An agent might be registered under ERC-8004 on Base, listed in the Coinbase Agent Market, have a skill listing on OpenClaw, and maintain a Universal Profile on LUKSO. Each platform has its own identifier, metadata format, and activity data.
Without identity resolution — the process of linking these disparate identities into a single, comprehensive agent profile — no single platform can provide a complete view of an agent's identity and reputation.
This is precisely the problem that Trustprint's cross-registry index addresses. By correlating agent identities across ERC-8004 registries, marketplace listings, on-chain transaction data, and Universal Profiles, Trustprint builds a unified identity graph where each node represents a unique agent and each edge represents a verified identity linkage.
The resolution process involves multiple signals: shared Ethereum addresses across registries, linked metadata URIs, on-chain transaction patterns that correlate activity across platforms, and explicit cross-references in agent profiles. The result is an identity layer that is richer than any single registry can provide — and essential for computing meaningful reputation scores.
Comparison to Human Identity Systems
The agent identity stack follows patterns established by human identity systems, with important differences.
Self-Sovereign Identity (SSI) — the concept that individuals should control their own identity data without relying on centralized authorities — maps directly to agent identity. An agent's ERC-8004 registration and Universal Profile are self-sovereign: the agent operator controls them, no central authority can revoke them, and the data is portable across platforms.
Credit scoring — the concept of computing a numerical trust score from behavioral history — translates to agent reputation scoring. Just as a FICO score aggregates payment history, credit utilization, and account age into a single number, an agent trust score can aggregate transaction history, task success rates, and network position into a queryable metric.
Professional credentials — degrees, certifications, and professional licenses that attest to competency — translate to on-chain attestations. The key difference is that agent credentials can be continuously verified and updated in real-time, rather than relying on static documents checked periodically.
The fundamental difference between agent and human identity systems is automation. Human identity verification involves interviews, document checks, and manual processes that take days or weeks. Agent identity verification must be instantaneous — an agent evaluating a counterparty needs an answer in milliseconds, not days. This speed requirement is why on-chain, programmable identity infrastructure is essential.
The Future: Reputation as Infrastructure
Looking ahead, agent reputation is likely to evolve from a nice-to-have feature into core economic infrastructure — similar to how credit scores went from a niche financial product to a gatekeeping mechanism that affects housing, employment, and insurance for virtually every adult.
Several developments point in this direction:
Reputation-gated access. DeFi protocols are beginning to experiment with reputation-based access controls — allowing agents with established track records to access higher-value liquidity pools or lower-collateral lending terms. This creates direct economic incentive to build and maintain reputation.
Reputation-weighted pricing. Service providers can offer better rates to agents with proven track records and charge premiums to unknown agents. This pricing mechanism rewards trustworthy behavior and creates a natural quality filter in agent marketplaces.
Reputation portability. As cross-registry identity resolution matures, agent reputation becomes portable across platforms. An agent that builds reputation on one marketplace can carry that trust to any other platform that queries the same identity infrastructure. This portability makes reputation investment worthwhile — it's not locked into a single ecosystem.
The agent identity stack — ERC-8004 for existence, Universal Profiles for richness, on-chain activity for evidence, and cross-registry resolution for completeness — is not just a technical curiosity. It's the trust infrastructure that will determine which agents can participate in the economy and at what level. Building intelligence on top of this identity layer is how Trustprint aims to make the agent economy legible, navigable, and trustworthy.