Very crisp description on the issues 2025+. What I would expect is OAth (OIDC4VC++) to start providing a bridge to at least more secure communication alternatives, such as replacing "bare metal" use of HTTPS with end2end secured protocols like ToIP's Trust Spanning Protocol, or adaptations of DIDComm. Certainly OIDC4VC is using DIDs and DIDComm in some scenarios for more secure communication.
In tech, the old hangs on with tweaks vs do-the-new commitment, until there is a catastrophic failure of the old, or the new finally overtakes the old via large organizations (public and private) finally moving on.
We are designing a communication protocol for agents called **ANP (Agent Network Protocol)**, which enables agents developed by any company to connect and collaborate over the internet. Our approach is quite different from A2A.
The most significant difference is that we use **DID (Decentralized Identifiers)** as the identity solution for agents. We believe DID is the most suitable scheme for agent identity.
Our open-source project has been running for more than a year (earlier than both MCP and A2A), and we are currently building an **open intelligent network**. You are welcome to follow us here:
Very crisp description on the issues 2025+. What I would expect is OAth (OIDC4VC++) to start providing a bridge to at least more secure communication alternatives, such as replacing "bare metal" use of HTTPS with end2end secured protocols like ToIP's Trust Spanning Protocol, or adaptations of DIDComm. Certainly OIDC4VC is using DIDs and DIDComm in some scenarios for more secure communication.
In tech, the old hangs on with tweaks vs do-the-new commitment, until there is a catastrophic failure of the old, or the new finally overtakes the old via large organizations (public and private) finally moving on.
Very helpful update for me, thanks
This is the best security-related article I’ve read in the past few months ❤️
This is an article I wrote about agent identity and authorization. I believe DID is more suitable for agents.
https://gaoweichang.substack.com/p/agent-identity-authorization-and
We are designing a communication protocol for agents called **ANP (Agent Network Protocol)**, which enables agents developed by any company to connect and collaborate over the internet. Our approach is quite different from A2A.
The most significant difference is that we use **DID (Decentralized Identifiers)** as the identity solution for agents. We believe DID is the most suitable scheme for agent identity.
Our open-source project has been running for more than a year (earlier than both MCP and A2A), and we are currently building an **open intelligent network**. You are welcome to follow us here:
🔗 [GitHub - Agent Network Protocol](https://github.com/agent-network-protocol/AgentNetworkProtocol)
In addition, here is our work at W3C:
🔗 [W3C AI Agent Protocol Community Group](https://www.w3.org/community/agentprotocol/)