Posts
All the articles I've posted.
- Updated:
AI Micro-Notes 2026: Thoughts Too Short to Trash
Short AI hot takes from 2026 onwards, accumulated from Threads and IG. Model roasts, dev pitfalls, industry observations, tool impressions — each no more than three lines.
-
The CLI Is the Firstborn
New features and fresh bug fixes land in the CLI first. Remote control waits forever. Same company, wildly different update cadence.
-
The Machines That Need the CLI Most Belong to People Least Likely to Learn It
For a machine short on resources the CLI uses a fraction of what a GUI does, but the people with the least headroom are also the least likely to ever learn it. Plus a note on how low the bar for local models actually is.
-
GPT-5.6 Luna Max Is the New King of Grunt Work
Using Luna Max for grunt work after the price cut: statements pulled in 22 minutes, brute-force scraping all night, and two ANA business class award seats from Bangkok to Tokyo.
-
The Skills and Hooks I Use Every Day, Now Open Source
I open-sourced the Claude Code skills and hooks I actually use every day. No technical skill required, pure logic, built for knowledge workers with no coding background.
-
A July Full of Resets: A Light AI User Reviews the Subsidy War
Every time a reset landed in July I used it to the last drop, and the ledger says 57.5x. Plus notes on the subsidy war, Deepseek's price hike, and my bet on the next reset.
-
Letting an Agent Tune a Local Video Model Overnight: My Three Gates
Three things I learned from letting an agent run local video-model parameter research overnight: hard constraints in CLAUDE.md, a gate script watching SSD writes, and a 30-minute check-in loop.
-
Installing Gemini CLI With Beginners Cures Most Claude Code Ailments
Five steps I walk Claude Code / Codex beginners through when installing the Gemini Antigravity CLI: multimodal use cases, getting comfortable with a CLI, a quick win from voice transcription, and finishing on data governance and local mode.
-
When to Call In Fable: Three Moments, and Don't Turn On ultracode
The most expensive model is not meant to run the whole way. Three moments where calling in Fable feels right to me, plus why I cap Opus 5 at med.
- Updated:
AI 碎念日記 2026：那些太短但捨不得丟的觀點
2026 年起在 Threads 和 IG 上累積的 AI 短碎念。模型吐槽、開發踩坑、行業觀察、工具心得，每則不超過三行。
-
CLI 是親生的，其他都是後媽養的
新功能和剛修好的 bug 都優先落在 CLI，remote control 要等猴年馬月。同一家公司的產品，更新頻率差很多。
-
最需要 CLI 的電腦，主人最不可能學會 CLI
對資源有限的電腦來說 CLI 佔用是 GUI 的零頭，但電腦資源最不夠的族群也剛好最不可能學會 CLI；順帶聊本機模型的硬體門檻其實沒那麼高。
-
GPT-5.6 Luna Max 是新一代的雜活之王
降價後拿 Luna Max 幹雜活：22 分鐘抓完各家報表、整夜土法煉鋼爬蟲、還幫我刷到 ANA 曼谷東京商務艙兩位。
-
我每天在用的 Skill 跟 Hook，開源了
把我每天都在用、不需要技術力、純邏輯導向的幾個 Claude Code Skill 跟 Hook 開源出來，最適合跟我一樣沒有編程背景的知識工作者。
-
充滿 reset 的七月：一個 AI 輕量用戶的補貼大戰復盤
七月一有 reset 就用好用滿，帳面發揮 57.5x 的價值；順便記下補貼大戰、Deepseek 漲價，和我對下一次重置的預測。
-
讓 Agent 整夜自己調本地影片模型：我設的三道閘門
讓 agent 夜間自己跑本地影片模型調參的三個經驗：硬約束寫進 CLAUDE.md、閘門腳本監控 SSD 寫入、30 分鐘回來盯一次避免快取失效。
-
帶初學者裝 Gemini CLI，治 Claude Code 百病
帶 Claude Code / Codex 初學者裝 Gemini Antigravity CLI 的五個步驟：從多模態 use case、CLI 介面的心理建設、語音轉錄的 quick win，一路帶到資料治理與本機模式。
-
何時派工 Fable？三個時機，還有別開 ultracode
最貴的模型不是拿來全程跑的。三個我體感最好的 Fable 派工時機，以及為什麼 Opus 5 我最多開到 med。
-
The Alignment Gap Is Closing. Next Comes Taste and Verification
The gap between AI and human intent is closing fast, so what separates good output moves toward taste and verification — and verification is where human responsibility stays.
-
Other People Fly to Korea for Cosmetic Surgery. I Had Codex Do Mine.
Someone on Twitter chained Codex, Hyperframes, IndexTTS2 and HeyGen into a one-person media pipeline. I pulled the repo, tried it, played with a local TTS model, then rebuilt a video days later on Claude Code.
-
Six New Context Engineering Rules for Claude 5, and the 1,473 Lines I Cut
After reading Anthropic's context engineering guidance for Claude 5, I turned the key points into nine cards, then cut 1,473 lines from a harness I had built up over four model generations.
-
Fewer Prompts Is Not Less Control
Claude 5 and GPT-5.6 official guidance is converging on the same thing — retiring old-style prompt stacking. But trimming is not letting go; control just moves. And the payoff of maintaining that layering is switching tools without losing a step.
-
My Local Model Lineup on a Mac mini, Plus Three Bad Habits I'm Owning Up To
Ten days of offline models on a 48GB Mac mini M4 Pro — the picks, the task split, a terrifying swap write rate, and three bad habits I admit to.
-
Porting Old Prompts to Opus 5: A Nine-Card Migration Guide
Nine cards I made after reading Anthropic's official Opus 5 prompting guide: response length, agent narration, task boundaries, subagent delegation, self-correction, and what happens when you turn thinking off.
-
Supposedly the Smartest Model, and Opus 5 Spent My Whole Day Spinning in Place
Wrong languages, sudden Simplified Chinese, then whole turns with no visible output — and a session log that matched an issue open since June 15.
-
Two Habits for Voice-Driving Coding Agents, and I Am Still Finding the Balance
Short instructions go through push-to-talk; walking through a whole course outline or system architecture goes through a full QuickTime recording that I hand to the AI to transcribe and execute. Plus the setup I use to burn the AI quota that came free with Google Drive.
-
對齊的 gap 正在縮小，接下來拚的是品味與驗證
AI 對齊人類意圖的 gap 正在快速縮小，區分產出品質的要素會往品味與驗證轉移；而驗證背後是人類永遠不會被取代的責任。
-
別人去韓國做醫美，我叫 Codex 數位醫美
從推特上看到有人把 Codex、Hyperframes、IndexTTS2、HeyGen 串成一條自媒體流水線，拉下來實測、順手玩 b 站的本地 TTS，幾天後換 Claude Code 重做一支。
-
Claude 5 時代的情境工程六條新規則，與我砍掉的 1473 行 harness
讀完 Anthropic 對 Claude 5 的情境工程指南後，我把重點做成九張圖卡，然後照著把累積四個世代的 harness 削掉 1473 行。
-
更少的 prompt，不是更少的控制
Claude 5 與 GPT-5.6 的官方指南正在合流，都在淘汰舊式 prompt 堆疊。但精簡不等於放任，控制只是換了位置——而維護好這套分層的紅利，是換哪個工具都能立刻接手。
-
Mac mini 上的本地模型陣容，順便招認我的三大劣根性
48GB 的 Mac mini M4 Pro 試了十來天離線模型，選型結論、任務分工、swap 嚇死人的寫入量，還有我承認的三個劣根性。
-
舊 prompt 搬到 Opus 5：九張卡的遷移指南
讀完 Anthropic 官方的 Opus 5 提示指南後整理的九張卡：回答長度、代理敘述、任務邊界、子代理委派、自我修正，還有關掉 thinking 的副作用。
-
號稱最聰明的 Opus 5，一整天在我對話裡空轉
從回錯語言、突然寫簡體，到整輪沒有可見輸出，查 session log 之後對上一個 6/15 就開著沒修的 issue。
-
語音下指令的兩種習慣，我還在找平衡
用 Claude Code 或 Codex 時，短指令我用隨按即錄，要盤點整套思路時我改開 QuickTime 完整錄音，再丟給 AI 轉錄執行。附上我拿 Google 雲端硬碟送的額度做轉錄的設定。
-
I'm Stuck at Stage 2.5 of AI Adoption
After watching Boris Cherny break down the stages of AI adoption, it hit home: I'm stuck at stage 2.5, where limited attention plus low trust becomes a vicious cycle. This week I used late-night schedules, a morning dashboard, and herdr auto-spawning sessions to cut a 4-5 hour workflow down to 1-2 hours.
-
Making Claude Talk Like a Human: A Plain-English Standard From Aircraft Maintenance Manuals
Claude stopped talking like a human starting with 4.7, and I thought my English just wasn't good enough — until Reddit showed me native speakers complaining too. So I turned ASD-STE100, the simplified-English standard used in aircraft maintenance manuals, into a skill, and now I can finally understand its English.
-
A Zero While Sitting on a Gold Mine
I read a piece on the perceived value behind churn-and-burn courses, turned it into a skill, ran it on my own site, and got back a verdict: a zero while sitting on a gold mine.
-
Some of My Subagents Were Already Grandpas
A CCX quota incident with no guardrails set: subagents bred recursively, one session burned 90% in half an hour, and here is the full forensics and fix.
-
Codex Built Its Own Evidence Package and Went to Argue With Google Support
A leaked Gemini backend key at PDT Learning got abused, no spending cap, and burned 1000 USD. I pointed Codex's browser automation at Google's live support to fight the charge, and it went so hard it built a 15-page evidence package and sent it over.
-
Two Traps Running Claude Code on Local Ollama: Truncated Context, and A3B Buckling Under Heavy Verification
Two things I logged: cco (Claude Code driven by local Ollama) had long been giving off-topic answers, and the root cause was not a weak model but a full harness whose system prompt had ballooned to 30-50k tokens and was being silently truncated; then I put A3B on the Mac mini for two days as a night worker.
-
Instead of Begging the Model Not to Lie, I Wrote a Hook That Stops It
The sequel to the Opus 4.8 confabulation post: I moved "don't make up numbers" from a plea in CLAUDE.md to a pending-guard hook that blocks git commit at PreToolUse.
-
Second Harness Diet: Global Skills From 58 Down to 40
A follow-up to the harness diet series: six months later, a bigger cleanup that cut my global skills from 58 down to 40, start to finish.
-
我卡在 AI 導入的 2.5 階段
看了 Boris Cherny 分享的 AI 導入階段框架後有感而發：我卡在 2.5 階段，注意力有限加信任不足變成惡性循環，這週靠深夜排程、晨間儀表板、herdr 自動開 session 把 4-5 小時的協作壓到 1-2 小時。
-
讓 Claude 說人話：一套飛機維修手冊的簡化英文標準
Claude 從 4.7 起就不說人話，我一度以為是自己英文太差，直到上 Reddit 才發現老外也在抱怨；後來我把飛機維修手冊在用的簡化英文標準 ASD-STE100 做成 skill，終於看得懂它的英文了。
-
坐擁金礦的零分
讀到一篇談割韭菜課程感知價值的文章，我馬上做成 skill 拿去盤點自己的網站，換來一句「坐擁金礦的零分」。
-
有些 subagent 都當阿公了
一次 CCX 沒設好護欄的額度事故：subagent 遞迴繁殖，一個 session 半小時燒掉 90%，事後鑑識與修法全記錄。
-
Codex 自己做了一份證據包，跑去跟 Google 真人客服吵架
PDT Learning 一支 Gemini backend key 外洩被盜刷、沒設 spending cap 怒噴 1000 USD，我用 Codex 的瀏覽器操作去跟 Google 真人客服爭費用，它認真到自己生出一份 15 頁證據包發給對方。
-
本機 Ollama 跑 Claude Code 的兩個坑：context 被截斷、A3B 撐不住重驗證
記錄兩件事：cco（本機 Ollama 驅動 Claude Code）長期答非所問，根因不是模型不夠聰明，是完整 harness 的 system prompt 早就膨脹到 3-5 萬 token 被靜默截斷；順手把 Mac mini 換上 A3B 當夜間工人測了兩天。
-
與其拜託模型不要騙我，不如寫一個擋得住的 hook
接續 Opus 4.8 捏造工具輸出那篇：我把「別亂編數字」從 CLAUDE.md 的拜託，升級成一個掛在 PreToolUse 擋 git commit 的 pending-guard hook。
-
第二次 harness 減肥：全域 skills 58 砍到 40
接續 harness 減肥系列，半年後做了第二輪更大規模的整理，把全域 skills 從 58 個砍到 40 個的完整過程。
-
A Few Days Into Switching From Claude Code to Codex: The Quota Honeymoon
A running log of switching from Claude Code to Codex this week: quota I could not burn through fast enough, a $20 Sol Ultra run that beat a $100 Fable plan on a professor friend's paper, and a tool-chain swap to Hyperframe.
-
從 Claude Code 換到 Codex 的這幾天：額度蜜月期
這週從 Claude Code 換到 Codex 的體感記錄：額度多到花不完、Sol Ultra 20 鎂幫教授跑論文，比 Fable 100 鎂方案還划算，工具鏈也跟著換成 Hyperframe。
-
Picking a Brand Look as a Non-Designer: AI Made Prototypes Cheap, So Taste Became the Hard Part
I am not a designer, but I spent this week iterating on a visual system for AgentCrew Academy with Fable, GPT-image-2, and GPT-5.6-sol. Here are the four rejected directions and the final spec applied to the website, slides, and documents.
-
外行人選品牌視覺：AI 把提案做便宜了，難的變成你的品味
我不是設計師，這週用 Fable、GPT-image-2、GPT-5.6-sol 迭代出 AgentCrew Academy 的視覺系統，記下被否決的四個方向與最後套用到網站、投影片、文件的規格。
-
Do Not Grade AI by Its Own Summary
I hit the same failure in several forms this week: an agent denied changing files, a commit message overstated the diff, and a transcript reassigned a speaker ID halfway through.
-
Four Things I Have Learned from Teaching AI
Recent corporate workshops reinforced four lessons for me: teach in person when possible, cut the slide count, show smart people the result first, and let students do the work.
-
I Finally Switched My Daily Driver from Claude Code to Codex
I try new tools easily, but I am slow to leave the ones already built into my routine. This time, quotas, pricing, and product direction all crossed the line together.
-
不要用 AI 的摘要驗收 AI
這週反覆踩到同一類錯誤：agent 說沒改檔卻留下檔案、commit message 與 diff 不符、逐字稿說話人中途漂移。驗收必須回到原始狀態。
-
最近教 AI 課學到的四件事
最近幾次企業培訓讓我重新確認四件事：實體課更好掌握節奏、投影片要做減法、聰明人先看結果，以及學員真正需要的是放手實作。
-
我終於把主力從 Claude Code 換成 Codex 了
我很愛試新工具，卻很難離開已經用習慣的工具。這次真正讓我換主力的，不是一次 benchmark，而是額度、價格與可預見的產品方向一起越過了臨界點。
-
I Let My /adhd Skill Go Dig Up a 3C-Scene Price-Hike Rumor on Its Own
I toss a vague rumor topic at my /adhd skill and watch it search, admit it searched wrong, stop to ask a disambiguating question, and finally piece the whole thing together with sources.
-
AI Micro-Notes 2026: Chronological Archive
The more scattered, time-sensitive AI micro-notes from 2026, in chronological order. Curated notes (by theme) live on the main page.
-
The Usage Economics of AI: Quotas, Tokenizers, and That Anesthetic Bill
A few scattered notes on AI usage, collected: pay-as-you-go vs subscription, a tokenizer quietly adding 1.4x, why resets are deliberately staggered, quotas getting pooled, and the bill you shouldn't mistake for output.
-
Fable 5's One-Week Return: Turning the Most Expensive Model into a Skill Distillation Engine
Fable 5 came back for one week only. I did not spend it on daily busywork — I spent it on high-leverage judgment, distilled into skills that cheap models can follow.
-
Sonnet 5 Is Out: The Lazy Version of the System Card, Plus Where I Actually Use It
Sonnet 5 shipped. I boiled the system card down to six plain-language points, then added where it actually fits: it is not here to fight Opus, it is here to replace Sonnet 4.6.
-
我讓 /adhd skill 自己去查一件 3C 圈的漲價八卦
含糊丟一個八卦題目給 /adhd skill，看它自己一路搜、承認搜錯、停下來反問消歧義，最後拼出全貌附來源。
-
AI 碎念日記 2026：時間軸存檔
2026 年較零碎、時效性的 AI 碎念，依時間排列。精選碎念（依主題分類）見主頁。
-
AI 的用量經濟學：額度、tokenizer，跟那張麻醉劑帳單
AI 用量的幾條觀察收攏成一篇：按量 vs 訂閱、tokenizer 悄悄多 1.4x、額度重置為何刻意打亂、獨立額度併池，還有那張別當成產能的帳單。
-
Fable 5 回歸一週：把最貴的模型變成 Skill Distillation 引擎
Fable 5 限時回歸一週，我沒拿去做日常小事，而是拿去做高槓桿判斷、蒸餾成便宜模型可照做的 skills。
-
Sonnet 5 發布：系統卡懶人版，跟我的實測定位
Sonnet 5 發布了，我把系統卡整理成六點白話懶人版，再加上實測定位：它不是來搶 Opus，是來換掉 Sonnet 4.6。
-
ADHD, Flow, and Why I Teach AI
Confessions of someone with ADHD: if I make it past two weeks and still have flow, odds are I can do this thing for a decade-plus.
-
The Client Says They Want Training, But Deep Down They Want a System
A post-mortem on letting go in a BD deal: the client drifted toward "give me an artifact" three times, and I changed one phrase from "after the training" to "if the priority is to build the hub" — and handed the ball back.
-
From Takedown to Return, I Caught a Case of AnthroPTSD
An observation diary of Anthropic models getting pulled by the government and apparently returning: from arrogance, to political wrangling, to OpenAI winning by default, to the conditioned reflex I've been trained into.
-
Claude Seeing Ghosts Four Nights Straight: A Log of Opus 4.8 Fabricating Tool Output
A four-day log of Opus 4.8 tool-result confabulation: from the technical symptoms to the GitHub issue to JSONL forensics.
-
ADHD、心流，與我為什麼教 AI
一個 ADHD 患者的自白：撐過兩週還保有心流，這件事我大概率能做十幾年。
-
客戶嘴上要培訓，骨子裡要系統
一次 BD 鬆手的覆盤：客戶三次飄向「給我 artifact」，我把一句話從「after the training」改成「if the priority is to build the hub」，球就交回去了。
-
從下架到回歸，我得了 AnthroPTSD
Anthropic 模型被政府下架到疑似回歸的觀察日記：從傲慢、政治角力、OpenAI 躺贏，到我被訓練出的條件反射。
-
Claude 半夜見鬼連續四天：Opus 4.8 捏造工具輸出實錄
Opus 4.8 連續四天的 tool-result confabulation 實錄：從技術現象、GitHub issue 到 JSONL 鑑識。
-
Opus 4.8 Cries 'Prompt Injection,' Codex GPT-5.5 Tracks Down the Real Cause: a Worktree Race
With multiple sessions open in one repo and no worktree, Opus 4.8 raised a false prompt injection alarm, told me to check for supply-chain attacks and rotate API keys, until Codex GPT-5.5 read the session log and pinned the real cause: just a worktree race.
-
Opus 4.8 搞烏龍報『提示注入攻擊』，Codex GPT-5.5 揪出 worktree race 真因
Opus 4.8 在我忘開 worktree 的情況下誤報「提示注入攻擊」，叫我去查供應鏈、rotate API key，最後 Codex GPT-5.5 查 session log 才揪出真因只是 worktree race。
-
Running Taiwan Sports-Lottery Odds With Mobile Claude Code, Standing in the Lottery Shop: the Odds Are Garbage, You'll Lose
Walked past a lottery shop, felt like placing a sports bet, and on a whim wired up a betting-odds skill on mobile Claude Code right there. The verdict: the correct-score market carries a 191% overround, every expected value is negative, you lose.
-
在彩券行用手機 Claude Code 算運彩賠率：台運賠率爛透了，誰下誰賠
路過彩券行起念想下運彩，順手用手機 Claude Code 接賭率分析 skill 現場開算，結論：台運波膽盤水錢高達 191%，期望值全負，誰下誰賠。
-
Which AI Is Strongest? I'll Take Them All: Claude as the Brain, Directing Gemini and Codex
Stop being a believer in one model being the strongest. Use multi-model collaboration: let Claude be the commanding brain and hand the right task to the right tool.
-
哪個 AI 最強？我全都要：讓 Claude 當大腦，指揮 Gemini 與 Codex
不要當「某家模型最強」的信徒，改用多模型協作：讓 Claude 當大腦指揮官，把對的任務交給對的工具。
-
No Error Doesn't Mean Success: Five Silent Failure Traps in AI Dev
A tool not throwing an error doesn't mean it succeeded. Exit 0, a 200 response, an empty string, a decapitated value — all silent failures. Reading back the real state is the only reliable defense.
-
沒拋錯不代表成功：AI 開發裡的五個靜默失敗陷阱
工具沒報錯不等於成功。exit 0、200 response、空字串、被砍頭的值，全都是靜默失敗——回讀驗證真實狀態才是唯一可靠的手段。
-
Two Mindsets for Getting Unstuck: Logical Thinking, and the AI Pedagogy of Never Making the Same Mistake Twice
Teaching a student today, I said that breaking through when you are stuck is not hard. What matters are two mindsets: logical thinking, and an AI pedagogy like raising a child, where the goal is not for the AI to never err, but to never make the same mistake twice.
-
卡關時的兩個心態：用邏輯思考，與『不貳過』的 AI 教育學
今天跟一個學員教學時談到，卡關時破除障礙並不難，重點是兩個心態：用邏輯思考，以及像帶孩子一樣的 AI 教育學，追求的不是讓 AI 無過，而是不貳過。
-
Teaching a Talent Agency to Use Claude Code, and the Soft Landing of the DK Confidence Curve
A short trip back to Taiwan packed with private training sessions: walking a talent agency through contract review, receipt reconciliation, and closeout reports with Claude Code, plus a note on the DK confidence curve and its soft landing.
-
帶藝人經紀公司用 Claude Code，與 DK 信心曲線的軟著陸
短暫回台跑私人場培訓，帶藝人經紀公司的企劃、製片、財務用 Claude Code 做合約審閱、單據核對、結案報告，順便聊聊 DK 信心曲線的快速上坡與軟著陸。
-
My First All-English Two-Hour Corporate Training: I Burned the Script Two Hours Before
Just wrapped the first session of an all-English corporate training for 30 people at an international hedge fund — my first time teaching two hours straight in English. I had a teleprompter and a full English script ready, then two hours before class I shut the teleprompter off and burned the script, and winged the whole thing.
-
第一次全英文兩小時企業內訓：上課前兩小時燒掉提詞稿
剛完成某國際 hedge fund 30 人企業內訓第一堂，這輩子第一次全程英文講兩小時。一開始準備了提詞機加英文全稿，上課前兩小時索性關掉提詞機、燒掉稿子，全程自由發揮。
-
The Perpetual Token-Burning Machine: Anthropic Really Is Trying
A few scattered gripes about Anthropic's relentless push to get users burning tokens: 103 auto-triggered agents, five-layer Russian-doll subagents, and Formosa-Plastics-tier flexing.
-
燒 token 永動機：Anthropic 真的很努力
幾則關於 Anthropic 在鼓勵用戶燒 token 這條路上的零碎吐槽：自動觸發 103 個 agent、五層俄羅斯娃娃 subagent、台塑等級的炫富。
-
The Five-Step SOP for Sneaking Claude Code Runs From the Hotel Over Dinner
Back in Taipei at a hotel, the five-step backup SOP I run before heading out to dinner with friends: pull power, keep it charged and online, caffeinate against sleep, tailscale, test remote desktop. Then sneak in some Claude Code over dinner.
-
出門吃飯前，讓 Claude Code 在旅館背景偷跑的五步 SOP
回台北入住旅館，出門跟朋友吃飯前的五步備援 SOP：取電、充電連網、防休眠、tailscale、遠端桌面試通。然後就能邊吃飯邊偷用 Claude Code。
-
From Miasma to Hades: How One Group Turned AI Tools Into a Supply-Chain Attack Vector
Two June 2026 npm/Python supply-chain attacks: Miasma backdoored Red Hat packages, then TeamPCP/UNC6780 upgraded to Hades, turning Claude Code, Cursor and 14 other AI tools into an attack vector. Includes self-check steps.
-
從 Miasma 到 Hades：同一個攻擊組織如何把 AI 工具當成供應鏈攻擊媒介
2026 年 6 月兩波 npm/Python 供應鏈攻擊：Miasma 入侵紅帽套件後門，TeamPCP/UNC6780 升級成 Hades，把 Claude Code、Cursor 等 14 款 AI 工具當成攻擊媒介。附自我檢查步驟。
-
Claude Fable 5 / Mythos 5 System Card: What's Actually Inside, and Why It Reads Like Sci-Fi
Flipping through the Claude Fable 5 and Mythos 5 system cards: stealing keys, office politics, AI infighting, gibberish, and a two-faced model. The more I read, the more it feels like a straight-faced sci-fi novel.
-
Testing Fable's Limits on a Professor Friend's Paper Repo
I woke up to Fei-bo (the guardrailed Mythos build) going live, so I borrowed a professor friend's paper proposal, prompt, and workflow docs to probe its paper-writing limits.