Post #3124448

Mention “workflow” in a prompt and Claude will dynamically create an orchestration plan that it strictly follows, allowing you to confidently trust that every stage happens in the right order even across 100s of agents. Previously, on Lobsters, we considered the degree to which Claude Code is configured via hard prompts instead of something more effective. Claude Code also often gets confused about its status in its internal workflow, the one which multiplexes chain-of-thought utterances (“thinking”), user input, and generated output (“confabulated bullshit”). Next time Claude Code source is leaked, I expect that we’ll see how poorly it “strictly follows” user-provided workflows, too.