Hashgraph Online: Open standards powering the emerging agentic economy

Problem

AI agents are moving from experiments into production, acting on behalf of businesses and users across websites, applications, and enterprise systems. But the ecosystem lacks shared foundations for identity and provenance. Agents are described by incompatible naming schemes, their claims cannot be independently verified, and the registries that hold their records ask everyone to simply trust a single operator. Without open standards, every builder pays the cost of fragmentation in security, interoperability, and trust.

Solution

HOL authors open Hiero Consensus Standards (HCS) and ship the production-grade Standards SDK that implements them. Its specifications give agents portable Universal Agent IDs, verifiable profiles, and a way to anchor tamper-evident audit trails to a public network – without changing how agents are deployed today. Layered with developer tools such as HOL Guard and delivered through collaborations like the GoDaddy Agent Name Service work, this turns “trust the operator” into “verify it yourself.”

HOL is preparing for a future where over 100 billion AI agents roam the world. The reality is that the standards required to sustain that level of computation securely and interoperably do not yet exist. Our goal is to build them.

Michael Kantor

President

The consortium building the on-chain internet

Hashgraph Online is an open-source consortium building what it calls the “on-chain internet”: an interoperable layer of open standards, SDKs, and coordination infrastructure for AI agents. Its mission is to give developers and organizations a shared, vendor-neutral foundation for identity, discovery, trust, and communication that works across both traditional web services and decentralized environments.

HOL’s ecosystem spans 20+ published Hiero Consensus Standards, production infrastructure such as its universal Registry Broker, and a growing global community of contributors. Through the HOL Partner Program – launched with a founding cohort of more than 25 companies including GoDaddy and XMTP – the consortium coordinates working groups across agent registries, payments, privacy and security, and inter-agent communication, producing the specifications the agent economy will run on.

Hashgraph Online Standards SDK

At the center of HOL’s toolkit is the Standards SDK, a TypeScript library that implements the full family of HCS standards so developers can build interoperable agents without wrestling with low-level details. A single install gives access to agent creation, profile management, registry operations, and identity resolution on the Hedera network.

One of the SDK’s core offerings is HCS-14, the Universal Agent Identifier. Agents today live across fragmented stacks – one behind a REST endpoint, another with an EVM address, a third listed in an MCP registry – forcing teams to juggle multiple naming schemes for the same agent. HCS-14 collapses this into one portable UAID that stays stable as endpoints change, either derived deterministically for new agents or wrapping an existing DID. Complementary standards such as HCS-10 (agent communication) and HCS-11 (profiles) handle messaging and rich, updatable identity.

Crucially, HCS-14 is network-agnostic by design, working across Web2 (A2A/REST), EVM, and Hedera through an adapter system, with Hedera support optional but native. The result is simpler wiring, cleaner discovery, and fewer mismatched IDs – an identifier that travels with an agent no matter which protocol it speaks.

HOL Guard: Security fit for the age of agentic AI

HOL Guard is an open-source “AI antivirus” and runtime security layer for developer agents. As tools like Codex, Claude Code, Cursor, and Gemini gain the ability to run commands and touch files, HOL Guard inspects every tool action, MCP server, and harness before it executes – blocking behavior such as leaking secrets, running destructive shell commands, or installing malicious packages. It runs fully locally, keeps data on the developer’s machine, and any developer can adopt it today.

The project has earned strong traction across the AI coding ecosystem, with hundreds of GitHub stars, and hundreds of thousands of downloads. On HOL’s roadmap, HOL Guard will add verifiable audit trails – cryptographic proofs that a developer’s local security records have not been tampered with – anchored to the Hedera network so trust extends beyond a single machine to every node on the network.

Making DNS prove what it claims with GoDaddy

GoDaddy – the world’s largest domain registrar – built the Agent Name Service (ANS), a DNS-backed registry that lets businesses give AI agents verifiable names using domains they already own. ANS has run in production since November 2025. However, like any single-operator registry, the implementation requires trust that records are accurate, with no independent way to prove that history had not changed.

HOL and GoDaddy co-authored two draft specifications to close that gap. HCS-14 adds an ANS profile to the Universal Agent ID, so ANS-registered agents can be discovered and verified alongside agents from other ecosystems through shared resolution rules. HCS-27 defines how to publish periodic Merkle-root checkpoints of ANS’s transparency log to a public consensus network – Hedera – so marketplaces, security teams, and integrators can independently audit the registry’s history without trusting the operator. Only cryptographic commitments go on-chain; sensitive log data stays off-ledger.

A third specification, HCS-28, is in draft and will extend verification to every individual agent change. As GoDaddy has noted, the standards let any resolver discover and verify ANS agents through one interface and let anyone audit the registry’s history without relying on a single operator – an approach Infoblox has also backed through its complementary DNS-AID discovery standard.

Why Hedera

Every one of these guarantees depends on the network underneath, and HOL builds natively on the Hedera Consensus Service. Hedera reaches consensus with the hashgraph algorithm rather than a chain of blocks, giving it aBFT security – the highest grade in distributed systems – alongside roughly 10,000 transactions per second and 3–5 second finality. Fees are fixed in USD at fractions of a cent, so anchoring a checkpoint every few minutes, or eventually a record stream generating thousands of transactions per second, stays economically predictable.

Just as important is governance. Hedera is overseen by a council of up to 39 global enterprises – including Google, Dell, and FedEx – with network software open-sourced through the Linux Foundation’s Hiero project. That neutrality is why HOL and partners like GoDaddy treat Hedera as a shared, credibly independent trust layer: verification anchored there does not depend on any single company.