XNews

Anthropic Publishes Guide on Core Primitives for Always-On Agents

Anthropic Publishes Guide on Core Primitives for Always-On Agents

Anthropic published a guide on building always-on agents, describing five core primitives: an event inbox, timers, memory, permissions, and an outbox. The guide explains that always-on agents are stateful and proactive, running on the same core loop as other agents but with surrounding architecture that controls when agents wake up, what carries over between runs, and how their work reaches users.

Event version 1.

Key insights

Five Core Primitives

The guide identifies five core primitives for always-on agents: an event inbox, timers, memory, permissions, and an outbox. These primitives control when agents wake up, what carries over between runs, and how their work reaches the user.

Stateful and Proactive Agents

Always-on agents are stateful and proactive, managing context across sessions and taking actions without direct supervision. They run on the same core loop as other agents, with the difference being the surrounding architecture.

Event Inbox for Multiple Triggers

Unlike most AI agents that only run when a user sends a message, always-on agents can be triggered by many types of events, including webhooks, scheduled tasks, and background jobs.

Run Cycle and Sleep Periods

An always-on agent spends most of its time asleep, waking up periodically to do work. Each stretch of work is called a run, during which the agent receives an event, completes its task, and saves information about what happened.

Source posts