MultiversX has just introduced the Agent Hub, a developer portal aimed at empowering AI coding agents within its blockchain ecosystem. Co-founder Beniamin Mincu hinted at this move in a tweet, stating, “We Are Setting Agents Free On MultiversX, Introducing Agentic Builder Hub, Game On.” This initiative represents a culmination of a year-long effort to elevate autonomous agents to reliable on-chain actors.
Traditionally, AI-assisted development promised to accelerate delivery by leveraging models to write or orchestrate code. MultiversX now argues that the landscape is evolving, bridging the gap between capability and discoverability. While agents capable of generating smart contracts or calling APIs exist, integrating them securely into a blockchain stack, selecting the right network, avoiding deprecated calls, managing denominations and gas, still requires human intervention. The Agent Hub provides the necessary context for agents in a machine-readable format, enabling them to “produce correct onchain code from the first prompt, not the fifth.”
The Agent Hub may seem simple in its ambition, but it boasts a complex infrastructure. It offers machine-readable discovery for agents to locate SDKs, workflows, and networks without manual setup. The platform incorporates behavioral guardrails within its documentation to prevent common errors, offers starter prompts for streamlined onboarding, and supports multiple agent runtimes. Compatible with Claude Code, Cursor, Windsurf, and Codex, builders have the flexibility to work with their preferred agent. MultiversX emphasizes that the Agent Hub complements, rather than replaces, the hands-on Builders Hub by catering to the agents relied upon by developers.
Beyond convenience, the Agent Hub serves as the gateway to a broader stack of protocols and standards introduced by MultiversX to facilitate agentic commerce. This stack includes MX-8004 for on-chain identity and interaction, UCP for agent-to-agent discovery, ACP/AP2 for commerce protocols, x402 for HTTP-native micropayments, and a MultiversX MCP server exposing 14 blockchain interaction tools directly to agents. Notably, MultiversX’s Relayed v3 enables gasless transactions, eliminating the need for agents or users to hold EGLD to initiate transactions.
The Strategy is Obvious
The goal is to seamlessly integrate identity, discovery, negotiation, and settlement into a unified developer experience, empowering agents to autonomously transact, negotiate, and settle value on-chain. MultiversX foresees economies where agents independently schedule services, reconcile invoices, procure data, and manage inventory with minimal human intervention. Commercially, the company emphasizes the importance of agents being able to easily access and utilize available tools within developers’ workspaces.
Developers can now outline requirements in Cursor or another agent-enabled IDE, allowing an agent to generate and deploy compliant smart contracts, call APIs, and handle micropayments without delving into extensive documentation. The Agent Hub aims to streamline the process for builders exploring autonomous agent economies, transitioning from hours of documentation to guided interactions. Early demonstrations, such as the “Max” agent, highlight the seamless workflow enabled by UCP discovery, AP2 delegated authority, ACP execution, and x402 settlement.
Questions surrounding security, governance, and the level of human oversight required will undoubtedly arise. MultiversX’s approach prioritizes setting constraints and machine-readable rules visible to agents, promoting safe defaults as the standard. Ultimately, the acceptance of cautious developers and regulators will hinge on real-world deployments and audits. Nevertheless, MultiversX is positioning itself as a leading testbed for agentic commerce. For those curious about exploring the Agent Hub firsthand, the opportunity awaits.