In the past, everyone probably thought that AI Agents wanting to do cross-chain operations had a higher barrier than humans.

You had to set up local deployments, study a bunch of scattered documentation, manually sign wallets. To be clear, it’s not the Agent doing the operations; it’s people working for the Agent.
But with this deBridge MCP upgrade, it’s not just adding a feature; it’s creating a true “connection layer.”
Previously, MCP was more geared toward developers and local environments. Now, it’s taking a step forward by supporting HTTP Streaming and natively connecting to browser Agents.
This means Agents no longer need to be confined to local environments—they can connect directly to cross-chain capabilities just like calling an API. Query routes, execute swaps, track transactions—all become basic actions.
More importantly, there are Agent Skills. In the past, many so-called “AI + DeFi” setups were essentially just data reading. Now, it’s becoming more fun—Agents can plan and execute on their own.
From “monitoring market trends” to “making strategies,” that crucial middle step has finally been filled.
And it’s no longer necessary to manually confirm each step—whether it’s same-chain operations or cross-chain asset management, everything is moving toward automation.
There’s also a detail I think is quite important: this time, the entry points have been unified. Landing Page + Remote-hosted Endpoint, essentially reducing duplication costs. Developers no longer need to cobble things together, and Agents can be integrated more quickly.
Especially with the Remote-hosted step, the use case has shifted from “development environment” to “real-world application.” Mobile devices and different AI clients can now connect easily.
Cross-chain with deBridge shouldn’t be a barrier; it should be infrastructure.
When Agents can query, trade, and transfer assets as easily as humans, multi-chain liquidity truly becomes accessible.
This is just the beginning. @deBridge_CN