Cluster Protocol x DopaMint: The Data and Builder Layer for Intent-Based Agents
Oct 1, 2026
4 min Read

An agent that understands what you want is only as useful as what it can reach. Ask one to size up a market and it needs data it does not already hold. Ask a developer to ship a new agent and they need somewhere to start that is not a blank repo.
DopaMint is building the intent side of that equation. Its House of Intent-Based Agents turns a plain-English request into coordinated work across 250+ specialized agents on Base. What those agents can pull from, and how quickly new ones get built, is where Cluster Protocol comes in.
Cluster Protocol and DopaMint are partnering on two fronts: connecting DopaMint agents to Cluster’s tokenized data marketplace, and bringing DopaMint-powered intent-agent templates into CodeXero.
Meet DopaMint
DopaMint🔗 is a House of Intent-Based Agents built on Base🔗. A user describes what they want in an iMessage-style chat. Dope, DopaMint’s orchestration layer, then works out which agents, tools and steps the job needs, from booking a restaurant to researching a market.
The roster is wide: 250+ specialized agents across markets and money, business operations, professional services, personal life, and creative and technical work.
Two design choices matter for this partnership. Agents act inside defined permissions, with a trust layer that runs deterministic policy checks on what each one is allowed to do. And payments run over x402, so an agent can pay for a service over plain HTTP as part of the job.
Where Cluster fits
Cluster Protocol is the orchestration layer underneath autonomous workflows. One endpoint sits in front of 500+ models for serverless inference, GPU compute billed per job, and a marketplace of tokenized datasets that owners list and price themselves. Payments run on x402 rails, and $CP is the settlement asset across the stack.
On top sits CodeXero, Cluster’s builder, where a developer goes from a prompt to a deployed app. CodeXero already deploys to Base, the chain DopaMint runs on.
So the split is clean. DopaMint has the users, the intent and the agents. Cluster has the data those agents consume and a place for the next ones to get built.
Tokenized data for DopaMint agents
The first piece connects DopaMint agents to Cluster’s tokenized data marketplace. Once it is live, an agent handling research, reasoning or a user-requested task can discover a relevant dataset on Cluster and access it as part of the job, without the user ever leaving the chat.
Datasets on the marketplace are listed and priced by the people who own them, and each purchase settles in $CP. For DopaMint, that means agents working from specialized data instead of whatever a general model happens to remember. For dataset owners, it means a new kind of buyer: an agent acting on a live request.

The request starts and ends in DopaMint’s chat. Only the data step runs on Cluster.
DopaMint intent-agent templates in CodeXero
The second piece runs the other direction. DopaMint-powered intent-agent templates will sit inside CodeXero, so a builder can start from a working agent pattern instead of an empty project. Pick the template, describe the app in plain English, and CodeXero builds it on DopaMint’s Agent Harness, which handles orchestration and execution.
Orchestration is the hard part of any agent app. Routing a request to the right tools and sequencing the steps is the code most teams get wrong on their first try. With the Agent Harness underneath, a builder spends their time on what the app does rather than on how its agents get coordinated.
The builder stays in CodeXero the whole way. The running app sits on DopaMint’s Agent Harness.
What each side gets
DopaMint’s agents get specialized data on demand, priced by whoever owns it, plus a second way for new agents to reach users through CodeXero. Builders get a starting point for intent-based apps that plugs into an orchestration and execution layer DopaMint already runs, instead of writing their own. Cluster gets demand from agents acting on real user requests, with each dataset purchase settling in $CP.
Nothing changes for the person in the chat. They keep typing what they want. The agents underneath just have more to work with.
What comes next
The first two pieces are the ones described above: dataset access for DopaMint agents, and the first DopaMint intent-agent template in CodeXero. Each one gets its own walkthrough when it ships, so builders can see the template end to end before they fork it.
If you build agents and want to be early on the template, follow @OpenProtocolFND🔗 and @DopaMintLabs🔗.
About Cluster Protocol
Cluster Protocol is the orchestration layer for autonomous workflows. One endpoint sits in front of 500+ models for serverless inference, GPU compute billed per job, and a tokenized data marketplace where owners list and price their datasets. Payments run on x402 rails, with $CP as the settlement asset across the stack. CodeXero, Cluster’s builder, sits on top and takes developers from a prompt to a deployed app. Cluster’s latest round, announced in April 2026, was led by DAO5 with Paper Ventures, JPEG Trading and Mapleblock Capital participating.
About DopaMint
DopaMint is a House of Intent-Based Agents built on Base. Through Dope, its orchestration layer, DopaMint connects users with 250+ specialized agents across markets and money, business operations, professional services, personal life, and creative and technical work.
