Summary

A feature tour of Claude’s lesser-known capabilities, each paired with a copy-paste prompt. It covers in-app features (Projects, Artifacts, Extended/Adaptive Thinking, Memory), role-based prompting personas (CBT therapist, hard mentor, personal trainer, devil’s advocate, conversation roleplay), and products (Claude in Chrome, Cowork, Scheduled Tasks, Skills, CLAUDE.md, Claude Code, Claude Design, Prompt Caching).

這是一份 Claude 鮮為人知功能的導覽,每項都附上可直接貼上的提示詞。內容涵蓋應用內功能(Projects、Artifacts、延伸思考、Memory)、角色化提示人格(CBT 治療師、嚴厲導師、私人教練、魔鬼代言人、對話角色扮演),以及產品(Chrome 版 Claude、Cowork、定時任務、Skills、CLAUDE.md、Claude Code、Claude Design、Prompt Caching)。

Key Points

  • Projects + Memory give Claude persistent context so you stop re-introducing yourself every chat.
  • Artifacts produce live, interactive apps (trackers, dashboards, games) inside the chat, available on the free plan.
  • Adaptive/Extended Thinking is off by default but materially improves answers on complex decisions.
  • Role prompts flip Claude from a validation machine into a challenger — the “hard mentor” and “devil’s advocate” prompts force counterarguments.
  • Product layer: Claude in Chrome acts on your active tab; Cowork accesses your filesystem; Scheduled Tasks run unattended; Skills install reusable capabilities; CLAUDE.md is read every session.
  • For developers: Prompt Caching cuts cost up to 90% on repeated context blocks via cache_control ephemeral.

Insights

The article’s framing — “most people have no idea” — points to a real adoption gap: the highest-value features (Memory, Thinking, Projects, role prompts) are off by default or undiscovered, so casual users experience a much weaker product than what is available. The unifying advice is to set up persistent context once and let it compound, rather than re-prompting from scratch each session.

Connections

Raw Excerpt

You now know more about Claude than most people who use it every day. Pick one feature from this list. Just one. Set it up today.