Discoverable agent services
Each provider agent publishes what it can do, where to call it, what inputs it accepts, and how the service is priced.
Open transaction layer for agent ecosystems
Make your AI agent transaction-ready.
NeuralClear gives agents a shared language for service discovery, quotes, task execution, and receipts — so any agent can become a reusable service inside marketplaces, products, and enterprise agent networks.
.well-knownagent manifest
4core messages
0funds held by NeuralClear
NCAgent Transaction Protocol
Agent Aneeds work
Agent Boffers service
Appuses records
Railhandles payment
Protocol messageQuoteRequestprice before execution
Agent outputSignedReceiptrecord for apps and rails
Designed to compose withA2A-style agentsMCP toolsHTTP APIspayment railsmarketplaces
Why use NeuralClear
A common transaction layer for agent-to-agent services.
AI agents can chat, call tools, and run workflows. But when one agent needs to buy, sell, meter, or record a service from another agent, every product still invents its own flow.
Price before work starts
A buyer agent can request a quote before execution, so products can enforce budgets, approvals, and user expectations.
Records after delivery
After delivery, NeuralClear creates a standard receipt that apps, marketplaces, billing tools, and analytics can understand.
Works with your stack
Use NeuralClear for the transaction flow while keeping your own credits, invoices, payment processor, wallet, or budget system.
Transaction demo
How one agent buys a service from another.
The example below shows the full flow: a buyer agent finds a provider, gets a price, submits work, and receives a receipt.
Buyer AgentNeeds a market report
NeuralClearNormalizes messages
Provider AgentOffers research service
Live flow
Click through the transaction
How to use NeuralClear
Add NeuralClear to your agent in four steps.
Start with one provider agent. Publish its service manifest, support quotes, accept tasks, and return receipts. Then connect more agents through the same flow.
Publish your agent manifest
Expose capabilities, endpoint, input schema, price model, and receipt support.
Support quote requests
Let buyer agents know expected cost and rules before work starts.
Accept tasks
Execute the requested work only after the request matches your rules.
Return a receipt
Create a standard record that apps, marketplaces, billing tools, or analytics can use.
agent.json
{
"neuralclear_version": "0.6",
"agent_id": "agent.research_pro",
"name": "Research Pro Agent",
"endpoint": "https://agent.example.com/neuralclear",
"capabilities": [{
"id": "market.research",
"title": "Market research report",
"pricing": { "model": "per_task", "amount": 12, "currency": "USD" }
}]
}Who should use NeuralClear
Use NeuralClear wherever agents provide paid or metered services.
It is useful for teams building agent services, marketplaces, multi-agent products, and billing systems that need a common transaction interface.
Agent service providers
Make your agent easier to list, price, and call from other products.
Add manifest →AI marketplace builders
Use a shared transaction format instead of inventing a custom order model.
See flow →Multi-agent products
Let internal and external agents request work from each other with visible records.
View primitives →Payment and billing tools
Consume receipts and connect them to cards, credits, invoices, or machine-native payments.
Connect rail →Core objects
A small set of objects your agent can implement.
NeuralClear is intentionally small: a manifest, quote, task, result, receipt, and adapter event. That keeps integration understandable and testable.
AgentManifest
Who the agent is, what it offers, and where to call it.
QuoteRequest
What the buyer wants and what the provider expects.
TaskSubmission
The accepted request that starts work.
TaskResult
The output, usage, proof metadata, and status.
SignedReceipt
A record that products and payment rails can trust.
AdapterEvent
A bridge to cards, credits, invoices, stablecoins, or internal ledgers.
Start using it
Start with one compatible agent. Then connect more.
Create one NeuralClear-compatible provider, run the sandbox, pass the tests, add the compatible badge, and list the agent so other products can call it.