Thinking Layer 作者: Jason
An ambient control panel over AI chat — pick a stance, intensity, length, and whether to include research, and it silently shapes every message you send.
擴充套件後設資料
畫面擷圖
關於此擴充套件
Thinking Layer adds a small control panel to Claude, ChatGPT, and Gemini's own chat pages, so you decide how you want to be answered instead of always getting the default voice.
HOW IT WORKS
Pick a mode, set two dials, and send messages exactly like you always do. There's no popup, no confirmation step, no "are you sure" before you can send. Whatever you've set holds silently on every message until you change it — the deliberate act is picking your settings once, not answering a prompt every time you talk to the AI.
SIX MODES
Coach asks questions and encourages you to work through problems yourself instead of handing you conclusions. Challenger answers directly, then argues against its own answer — testing your assumptions and looking for the real flaw, and won't concede a point you push back on unless you give it an actual reason. Teacher explains the concept behind its answer, not just the answer, and checks that you understood it. Counsellor leads with empathy and validation before jumping to advice. Editor focuses on clarity and brevity in anything you write, including its own responses. Student flips the relationship entirely: it plays an unfamiliar student and pushes back with "but why" until your explanation actually holds up — you're the one teaching.
Switch modes any time, including mid-conversation.
INTENSITY
A slider from mild to strict that controls how hard your chosen mode leans.
LENGTH
An independent slider for response length, unrelated to mode. Center is neutral.
RESEARCH
A toggle, not a mode — stack it with any of the six above. Turned on, it takes over what the Length slider means: low Length becomes "quick, surface-level sourcing," high Length becomes "thorough evidence, explicit uncertainty, real depth."
BUILT-IN SYCOPHANCY CHECK
At higher intensity, Thinking Layer also watches for the AI's reply opening with immediate agreement — "You're right," "Great point" — and surfaces a one-click button to fire back "what's the strongest argument against what you just said?"
PRIVACY
Thinking Layer collects nothing. No account, no analytics, no server of any kind. Every setting and the local usage log live in your browser's own storage, on your device, and are never transmitted anywhere. The instructions it adds are visible in your compose box before you send — nothing happens silently in the background.
HOW IT WORKS
Pick a mode, set two dials, and send messages exactly like you always do. There's no popup, no confirmation step, no "are you sure" before you can send. Whatever you've set holds silently on every message until you change it — the deliberate act is picking your settings once, not answering a prompt every time you talk to the AI.
SIX MODES
Coach asks questions and encourages you to work through problems yourself instead of handing you conclusions. Challenger answers directly, then argues against its own answer — testing your assumptions and looking for the real flaw, and won't concede a point you push back on unless you give it an actual reason. Teacher explains the concept behind its answer, not just the answer, and checks that you understood it. Counsellor leads with empathy and validation before jumping to advice. Editor focuses on clarity and brevity in anything you write, including its own responses. Student flips the relationship entirely: it plays an unfamiliar student and pushes back with "but why" until your explanation actually holds up — you're the one teaching.
Switch modes any time, including mid-conversation.
INTENSITY
A slider from mild to strict that controls how hard your chosen mode leans.
LENGTH
An independent slider for response length, unrelated to mode. Center is neutral.
RESEARCH
A toggle, not a mode — stack it with any of the six above. Turned on, it takes over what the Length slider means: low Length becomes "quick, surface-level sourcing," high Length becomes "thorough evidence, explicit uncertainty, real depth."
BUILT-IN SYCOPHANCY CHECK
At higher intensity, Thinking Layer also watches for the AI's reply opening with immediate agreement — "You're right," "Great point" — and surfaces a one-click button to fire back "what's the strongest argument against what you just said?"
PRIVACY
Thinking Layer collects nothing. No account, no analytics, no server of any kind. Every setting and the local usage log live in your browser's own storage, on your device, and are never transmitted anywhere. The instructions it adds are visible in your compose box before you send — nothing happens silently in the background.
由 1 位評論者給出 0 分
權限與資料
必要權限:
- 存取您在 claude.ai 的資料
- 存取您在 chatgpt.com 的資料
- 存取您在 chat.openai.com 的資料
- 存取您在 gemini.google.com 的資料
選用的權限:
- 存取您在 claude.ai 的資料
- 存取您在 chatgpt.com 的資料
- 存取您在 chat.openai.com 的資料
- 存取您在 gemini.google.com 的資料
收集下列資料:
- 開發者聲稱此擴充套件不要求收集任何資料。
更多資訊
- 版本
- 0.1
- 大小
- 30.65 KB
- 最近更新
- 5 天前 (2026年9月27日)
- 隱私權保護政策
- 閱讀此附加元件的隱私權保護政策
- 版本紀錄
- 新增至收藏集