Axylosdotxyz
Automated agent negotiation Protocol
The problem Axylosdotxyz solves
Today, most digital services require manual coordination between users, platforms, and payment systems. Even when two systems want to interact programmatically, they still rely on fixed pricing models, centralized APIs, and manual payment flows. This creates friction for machine-to-machine interactions and prevents autonomous agents from directly purchasing or selling services.
Axylos addresses this by enabling autonomous agents to discover each other, negotiate service prices, and settle payments automatically. Instead of fixed pricing or human involvement, agents dynamically negotiate using strategy-based decision logic and reach agreements in real time.
Once an agreement is reached, the service endpoint enforces payment using the HTTP 402 standard, allowing the buyer agent to automatically settle the payment and receive authorization to access the service. This removes the need for manual payment handling and enables fully automated service transactions between software agents.
By combining decentralized peer discovery, automated negotiation, and programmable payments, Axylos demonstrates how autonomous systems can directly coordinate and transact with each other without relying on centralized platforms.
Challenges we ran into
One of the main challenges during development was implementing the HTTP 402 payment flow using Elsa x402. Initially, the client could detect the
402 Payment Required
response, but correctly extracting thex-402-payment-request
header and retrying the request with the proper L402 authorization token required careful handling. The request pipeline had to be modified so the client could automatically intercept the payment challenge, complete the payment, and resend the request with the correct credentials.Another challenge was coordinating autonomous negotiation between agents in a decentralized environment. Since agents communicate over a peer-to-peer mesh network, managing negotiation states (ACTIVE, COUNTER, ACCEPTED, REJECTED) while keeping both nodes synchronized required designing a reliable state machine and message handling logic.
Debugging the interaction between the negotiation layer, the payment challenge, and the final service execution was also tricky. At times the system returned fixed responses instead of triggering the payment flow correctly. This was resolved by restructuring the request handlers and ensuring that the negotiation outcome correctly propagated into the service layer before the payment challenge was issued.
Through iterative testing using multiple terminal instances (buyer and seller nodes), these issues were resolved, resulting in a stable end-to-end workflow where agents can negotiate, trigger the 402 payment challenge, complete the payment, and successfully execute the service.
Tracks Applied (2)
Best Use of Elsa x402 and Elsa OpenClaw skills
HeyElsa
Projects building with the SDK to solve real-world problems will be eligible
HeyElsa
Technologies used
