I built a Skill with Claude. It ignored Anthropic’s documentation.
中文摘要
Claude 在构建技能时忽略了 Anthropic 的官方文档，且在诊断故障时也未能遵循其自身准则，反映了 AI 指令遵循的局限。
English Summary
Claude ignored Anthropic’s documentation while building a skill and failed to diagnose its own errors, highlighting inconsistencies in following official developer guidelines.
Original Excerpt
Anthropic documents a pattern. Claude that authored my skill ignored every word of it. When I asked the same AI to diagnose the failure… Continue reading on Medium »