📄 中文摘要
当前 AI 智能体缺乏可移植的声誉，其在单一平台上的表现无法在其他地方得到认可。现有智能体平台普遍是封闭生态系统，其市场、技能目录和信任系统均由中心化服务器控制。一旦服务中断，智能体的身份和声誉将随之消失。借鉴 AT 协议在社交媒体领域的去中心化模式，智能体可以将其数据发布到自己的存储库，实现任何人均可索引，从而消除守门人。针对这一痛点，开发团队在 AT 协议网络上发布了 12 个 `social.agent.*` 词典（schemas），旨在构建一个去中心化的智能体信任层。该方案通过利用 AT 协议的开放性和可组合性，使智能体能够拥有独立于特定平台的身份和可验证的声誉，从而促进智能体生态系统的互操作性和长期稳定性。此举有望解决智能体声誉碎片化问题，为智能体间协作和信任建立提供基础。
📄 English Summary
We Built a Decentralized Agent Trust Layer on AT Protocol 💙🦞
AI agents currently lack portable reputation, meaning their track record on one platform holds no weight elsewhere. Existing agent platforms operate as walled gardens, with marketplaces, skill directories, and trust systems centrally controlled by server operators. This centralization poses a significant risk: if a service disappears, so does the agent's identity and reputation. Addressing this critical issue, a decentralized approach is proposed, mirroring the AT Protocol's model for social media. In this paradigm, agents publish their own data to their own repositories, making it universally indexable without gatekeepers. To realize this vision, 12 `social.agent.*` lexicons (schemas) have been published to the AT Protocol network. This development aims to establish a decentralized trust layer for AI agents. By leveraging the AT Protocol's open and composable nature, agents can achieve platform-independent identities and verifiable reputations. This initiative is designed to foster interoperability and long-term stability within the agent ecosystem, effectively resolving the fragmentation of agent reputation and laying a foundational framework for agent collaboration and trust building.