Anthropic launches memory for Claude Managed Agents in public beta

Anthropic 今天为 Claude Managed Agents（托管智能体）上线了记忆功能，目前处于公测阶段。简单说，你部署的 AI 智能体现在能在多次会话之间积累经验，下次干活时不用从零开始。

Anthropic has launched memory for Claude Managed Agents in public beta, letting agents retain and reuse experience across sessions. Memory is stored as files on a filesystem, with shared permissions, concurrent access, audit logs, and rollback; Rakuten reports a 97% drop in first-time errors, and Wisedocs reports 30% faster document validation. The key detail is the implementation path: it uses a filesystem, not a dedicated vector database.

Why it matters: Anthropic adds cross-session memory to Claude Managed Agents beta and discloses the implementation plus two user numbers: Rakuten 97% and Wisedocs 30%. HKR-H/K/R all pass, but the scope is still limited to the managed-agent beta, so this lands at 83 and featured.