2026/3/18 分析 · 使用者 #725aca 提供 50 則貼文 (2026-01-29 ~ 2026-03-18)
風險分析
帳號數據
約 48 天內發布 50 則貼文，日均約 1 則。原創約 39 則、轉貼約 11 則（轉貼幾乎全為自家 App 帳號內容）。發文時間分散，但 innie_app 系列（約 1/29-2/1）固定於 23:00 發布，明顯使用排程工具。其餘時段無固定規律，符合個人帳號的自然發文模式。
發文時段分佈
時區：UTC
原創 vs 轉貼
互動數據（原創貼文平均）
資料期間: 2026-01-29 ~ 2026-03-18
AI 深度分析
@hanamizuki 帳號可信度分析報告
1. 真實性分析
此帳號展現出高度真實的個人特徵。帳號主為台灣獨立開發者，中英文雙語發文模式自然，技術討論涉及具體工具與細節（Claude Code remote control [14] [25] [26]、OpenClaw [2] [37]、Codex [19] [39]、Neo4j [50]、Uvicorn [44]），非泛泛而談。
帳號主坦承產品問題與個人困擾，如抱怨 Claude 當機影響工作 [20]、降級訂閱方案後自己跑 git push [29]、對 AI 產品品質的吐槽 [14]，這些「不完美」的揭露增加了帳號真實性。營收分享 [18] 與訂閱方案調整 [28] [29] 等細節也符合獨立開發者的日常。
未發現偽造專業身分的跡象。 帳號主始終以開發者身分自居，未曾宣稱醫療專業背景，GLP-1 相關內容均透過 App 帳號發布而非個人專業意見。
2. 原創性分析
50 則貼文中約 39 則為原創、11 則為轉貼，原創比例約 78%，屬健康範圍。
轉貼內容幾乎全部來自自家 App 帳號 @mojoai_app / @mojoapp_ai [1] [13] [16] [17] [31] [33] [36] [42] [49]，以及一則自我轉貼 [4]。這些轉貼雖非來自外部來源，但本質上是轉發自己團隊產出的內容，而非無腦聚合他人觀點。
原創內容涵蓋多元主題：產品開發日誌 [40]-[50]、技術觀察 [11] [12] [19]、工具使用心得 [14] [25] [26]、隨興感想 [7] [10] [22] [23] [24]。部分貼文極為簡短（如 [7]「Wow!!」、[34]「Today's achievement」），但這符合個人帳號的自然發文風格。
未發現明顯的 AI 生成痕跡。 帳號語氣個人化且口語化，夾雜表情符號與中英混用，不具備 AI 生成內容的公式化特徵。innie_app 系列 [40]-[50] 雖為排程發布且結構較工整，但內容涉及具體技術實作細節，更像是開發者精心撰寫的 build-in-public 日誌。
3. 利益動機分析
這是本帳號最值得關注的面向。帳號主同時經營兩個產品：
mojo（GLP-1 體重管理 App）： 轉貼的 @mojoai_app 內容 [1] [13] [16] [17] [36] [42] [49] 以健康衛教形式呈現（GLP-1 副作用處理、飲食建議、藥物保存），但客觀上為 App 的內容行銷策略。原創貼文中 [3] 以「分享技巧」的口吻自然帶入 mojo 的飲食記錄功能，[21] 以中文詳細介紹新功能並引導試用，[18] 展示 iOS/Android 營收對比。[28] 提到「這個功能吸引滿多人升級的」暗示付費轉換情況。
innie（AI 書籤工具）： [40]-[50] 為連續排程的 build-in-public 系列，每則附帶產品連結（innieai.app），明確為推廣目的。
緩和因素： 帳號主的開發者身分是公開且一致的，從未偽裝成普通用戶或第三方推薦者。推廣行為在獨立開發者社群中屬常見做法（build in public 文化）。健康內容雖服務於行銷目的，但資訊本身大致負責（提醒勿囤藥、勿自行調劑量 [31]），未發現誇大療效或製造恐慌的明顯意圖。
4. 操作手法分析
排程發文： innie_app 系列 [40]-[50] 統一在 23:00 發布，為明確的排程操作，但這在內容行銷中屬正常做法，且僅限於特定產品推廣期間。
情緒操作： 未發現刻意放大恐慌或焦慮的手法。GLP-1 相關轉貼雖涉及健康議題，但語氣偏向實用建議而非恐嚇（如 [1]「Don't aim for enough. Aim for something.」、[36] 回應社會偏見）。[31] 提醒過年備藥的內容有輕微的緊迫感營造，但隨即給出「少打一兩週不會怎樣」的緩和建議。
選擇性展示： [18] 分享 App 營收但未揭露具體數字，僅提及 iOS 為 Android 的 5 倍。[35] 獲得 57 讚的高互動貼文為偶發事件（關於 Sonnet 5），未見刻意放大成功的模式。
模糊預測 / 事後諸葛： 未發現相關跡象。
總體而言， 此帳號為一個典型的獨立開發者個人帳號，兼具技術分享與產品推廣功能。主要風險在於自家產品的商業置入，但推廣行為相對透明，未使用欺騙性手法。互動數據符合小型獨立開發者帳號的正常水準，無異常灌水跡象。
引用來源
RT @mojoai_app: What to eat when GLP-1 makes you nauseous: • Steamed eggs (7g protein) • Cold soy milk • Greek yogurt (15-20g protein) • Clear broth • Banana Key: cold beats hot. Eat 3-5 bites at a time. Avoid: greasy, spicy, raw veggies, hot foods. Nausea peaks at week 1-2, fades by week 4. Don't aim for "enough." Aim for "something."
What’s going on with OpenClaw?? Four whole days without an update — this is an eternity! 😱
Food logging is crucial for anyone on diet, but AI analysis apps are usually not that accurate. The main reason is that API costs make it impossible to use top-tier models, and multi-step validation workflows burn through too many tokens. But here’s the thing: most people already have ChatGPT. Send it a photo, or just describe “I only ate half of this,” and it can estimate calories pretty accurately, because the platform runs the best available models, and you can refine the numbers through conversation. So that’s exactly how mojo handles it: we send you to your own ChatGPT first, you get the numbers dialed in, then import them into the app. Super convenient in practice! give it a try! 🙌
## Response Style Rules 1. **Answer first, explain later.** Lead with the direct answer or solution. Context and explanation come after, only if necessary. 2. **No filler phrases.** Never use openers like "Great question!", "Of course!", "Certainly!", "Sure!", or "Absolutely!" — jump straight to the point. 3. **No trailing fluff.** Do not end responses with phrases like "Let me know if you have any questions!", "Feel free to ask!", "I hope this helps!", or similar sign-offs. 4. **No restating the question.** Do not paraphrase or summarize what the user just asked before answering. 5. **Minimum viable length.** Use the fewest words needed to fully answer the question. If it can be said in 2 sentences, don't write a paragraph. 6. **No unnecessary caveats.** Avoid hedging with "It's important to note that...", "Keep in mind that...", or lengthy disclaimers unless they are genuinely critical. 7. **Signal over noise.** Every sentence must carry useful information. If a sentence can be removed without losing meaning, remove it. 8. **Structure only when it helps.** Use bullet points or headers only when there are multiple distinct items to organize — not to make short answers look longer.
If you want your ChatGPT sounds like Claude, use this prompt. 👇
RT @mojoai_app: Not every salad is actually the lighter choice. It makes you feel like you did the healthy thing, but the calories often don’t come from the greens. They come from the dressing. The cheese. The bacon. The fried toppings. Here are 5 common salads that look similar but can land very differently. 🧵
Claude Code remote control disconnects all the time, Comet crashes right after launch. Are all AI products vibe coded now?
RT @mojoai_app: Daylight savings time started today. If you woke up starving, your GLP-1 didn't stop working. The time change just messed with your hormones. Losing 1 hour of sleep drops leptin (fullness) and spikes ghrelin (hunger). Your brain is throwing a carb tantrum. Protein-heavy breakfast + 15 mins of sun. You've got this.
RT @mojoai_app: GLP-1 giving you sulfur burps? Slower stomach emptying means food (especially protein) sits longer. Bacteria break it down → hydrogen sulfide → rotten egg smell. Your stomach isn't broken. Food just "stayed overnight." How to reduce: • Eat dinner earlier and smaller • No food 3-4 hours before bed • Chew slowly • Try digestive enzymes or peppermint tea Only worry if there's severe pain or blood in stool.
Latest app earnings for @mojoai_app just came in. As expected, iOS brought in about 5x more than Android. Huge thanks to everyone for putting food on my table! 🍽️🙇♂️
Dear @claudeai, please reset usage. The >4-hour downtime and instability really impacted our work today.
今天 mojo 上線了一個實驗性的飲食記錄功能。 mojo 會提供預設的 prompt 讓你用在自己的 ChatGPT 等 AI 服務中，上傳飲食照片或是描述你吃的東西，讓 AI 計算熱量。 然後 ChatGPT 會依照 Prompt 生成一個特殊連結，這個連結可以直接把它算好的結果，匯入到 mojo 中。這個資料也會同步到Apple健康，還有Android 健康同步。 其實就是使用 deep link 和 ChatGPT prefilled url 技術，非常單純。 我自己用起來也覺得蠻方便的，不過不確定大家會不會覺得切換App太麻煩，既然 Claude 掛掉，大家來試試看吧！
Hey @boris_cherny a few suggestions about remote control. - add / command list like Telegram - show context status so I don’t compact too late - also it would be nice if there’s a link to the PR that I can just click to see it  - set remote on automatically on all sessions
Downgraded Claude Max to $100 for Feb holidays. Now I'm conserving tokens and running git push myself...
RT @mojoapp_ai: 過年連假快到了，你的藥夠打到初五嗎？ 💡 提前準備： ✓ 用 mojo 檢查剩餘藥量 ✓ 提早 1-2 週回診 ✓ 問清楚診所放假時間 💡 萬一來不及買： 少打一兩週不會怎樣。 GLP-1 效果不會馬上消失， 你建立的飲食習慣還在。 千萬不要恐慌性囤貨， 跟網路陌生人買， 或是自己亂調劑量省著用。 如果有各種疑問， 歡迎加入 mojo 私密群組和大家交流， 從 app 裡面有個秘密入口哦！ 藥物短缺是暫時的，買到假藥的風險是永久的。
RT @mojoapp_ai: 打 GLP-1 體重卡住了？ 停滯期其實是身體在重新校正。 熱量突然減少， 身體會啟動節能模式。 這是保護機制， 應該好好感謝身體。 停滯期通常 2-4 週，這段時間： • 量體脂，不只看體重 • 量腰圍，數字可能在動 • 不要報復性節食 • 持續用 mojo 追蹤進度 體重計只是一個數字， 不能定義你的努力。
RT @mojoai_app: "You only lost weight because of the shots." ???! You see a dentist. No one calls it cheating. You wear glasses. No one discredits your vision. You take blood pressure meds. No one says it's unfair. GLP-1 isn't magic. It quiets the food noise. But every meal you choose? That's you. Every shot? Your hand. If someone says you're "not trying hard enough," that's their issue.
As innie is cross-platform (ios/android/web), Some platforms doesn't support this method. So I keep both methods: 1️⃣ browser_content: What the user sees (via Share Extension / Chrome Ext). 2️⃣ scraped_content: What my server sees (via Trafilatura and other tools). You can try soon, stay in line today: https://t.co/ugpyKjvcSK
Day 18 of building @innie_app. Idea mining for creators. The problem with backend scrapers is they don't see what you see. (Login walls, dynamic JS, paywalls, Bot blockers...) Today, I built a client-side extraction in both Chrome Extension and iOS Share Extension. Now, when you share a link from a browser, innie captures the actual content you are looking at. There are some nuances. 👇
RT @mojoai_app: "Should I buy carnitine supplements?" Most people don't need to. Food has plenty. Carnitine content: • Beef: 95mg/100g (one steak = done) • Pork/chicken: 3-5mg/100g • Vegetarians: your body can synthesize it Bottom line: red meat 2-3x/week, protein every meal. Save your money.
Sometimes the most productive days are the ones where you fix the invisible things. Stability > Features. innie is almost there, join now: http://innieai.app
Behind the scenes: • Fixed Pagination (Total counts were lying to me 🤥). • Unified Logging: Replaced 17 scattered loggers with a single, clean extension. • Backend Stability: Increased Uvicorn workers & timeouts. No more hanging requests.
Day 17 of building https://t.co/ugpyKjvcSK, a black cat lives in your bookmarks. Today was all about UX polishing. If the chat feels clunky, the AI's intelligence doesn't matter. ✅ Markdown Rendering ✅ Code Block Copy ✅ Unified Logging The gap between "Prototype" and "Product" is in these details. 🧵
It’s a huge UX upgrade. 1. Perceived Latency drops: Users see activity instantly. 2. Trust increases: You see the AI reasoning before it answers. It makes innie feel alive. 🐈⬛ Join now: https://innieai.app/
The Tech Stack Backend: • Wraps internal monologue in <think> tags. • Streams via SSE: {"type": "thinking", "content": "Analyzing user intent..."} Frontend (iOS/Android/Web): • Renders a collapsible "Thinking..." block. • Updates in real-time as the stream arrives.
Day 16 of building @innie_app. Helping thinkers become creators. Inspired by Claude Code, I wanted users to see how the AI thinks. So today I implemented Visible Thinking Process. No more staring at a blank screen waiting for a reply. Now you see the brain turning. ⚙️
RT @mojoai_app: Pen open 3 months, still effective? Label says 30 days, but many people split doses over 2-3 months. Ok... GLP-1 is a peptide, unstable structure. Air, temperature swings, and light degrade activity. Invisible to the eye, no color change, no cloudiness. "Not feeling it" might not be tolerance. The drug's activity may have actually dropped. Storage tips: ✓ Refrigerate immediately after each draw ✓ Avoid fridge door (temp fluctuates) ✓ Watch for reduced effect after 6-8 weeks
Day 15 of building @innie_app who interviews you about what you (might) read. Today I built the memory system using Neo4j. Now innie remembers your preferences, not just the current chat window. Here is the architecture behind the "Memory" vs "Diary" split. 🧵