https://zheng-bobo.github.io/en/tags/anthropic-agent-evolution/ Recent content in Anthropic Agent Evolution on Blowing in the wind Hugo -- gohugo.io en Sat, 12 Sep 2026 21:08:45 +0200 https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-1-from-workflows-to-agents/ Sat, 12 Sep 2026 21:08:45 +0200 https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-1-from-workflows-to-agents/ <p>“Agent” is often used as a broad label for any system that calls tools or takes multiple steps. Anthropic&rsquo;s engineering work offers a more useful distinction: <strong>a workflow follows code paths defined in advance; an agent lets the model dynamically choose its process from environmental feedback.</strong></p> <p>This first article in the “Anthropic Agent Evolution” series establishes the vocabulary: augmented LLMs, fixed workflows, autonomous agents, and the harness that surrounds the model.</p> https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-2-context-reset-and-handoff/ Sat, 12 Sep 2026 21:07:45 +0200 https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-2-context-reset-and-handoff/ <p>The central tension in long-running work is not whether a model can write code. It is that a task may last hours or days while every context window is finite. Replaying the entire history eventually creates both capacity and noise problems.</p> <p>Anthropic&rsquo;s first long-running agent harness answered with a simple principle: <strong>reset conversational context, but externalize project state into structured artifacts that the next agent can recover.</strong></p> https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-3-planner-generator-evaluator/ Sat, 12 Sep 2026 21:06:45 +0200 https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-3-planner-generator-evaluator/ <p>The first harness let an agent continue across context windows, but continuity is not quality. Generators routinely mistake “mostly runs” for “good enough,” especially in subjective product and design work.</p> <p>Anthropic&rsquo;s next step separated planning, generation, and evaluation so that an external evaluator became a source of evidence the generator had to respond to.</p> https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-4-managed-agent-runtime/ Sat, 12 Sep 2026 21:05:45 +0200 https://zheng-bobo.github.io/en/post/anthropic-agent-evolution-4-managed-agent-runtime/ <p>Earlier harnesses focused on how a model could continue work and improve quality. A managed service introduces system questions: how is a session persisted, how does work recover after a harness crash, how can sandboxes be replaced, how do customer VPCs connect, and how are credentials kept away from model-generated code?</p> <p>Anthropic Managed Agents answers by <strong>decoupling the brain, hands, and session behind stable interfaces.</strong></p>