← The AI Hype Audit — all 272 verdicts

PARTLY Partly true: the 30-minute Claude Cowork folder setup is real and free, but the script is a lift of a March post with the model bumped to a legacy 'Opus 4.8', and the 'comment CLAUDE' guide has no trail Facebook · Sep 2026

The claim10-second reel, ~9.6K views, sent Sun Sep 6, 2026: 'Spend 30 minutes on this Claude setup and every answer stops sounding generic.' Recipe in the caption: open Cowork instead of chat, turn Extended Thinking on, select Opus 4.8, point Claude at a 'Claude Work' folder with About Me / Projects / Templates / Outputs, add about-me.md, my-voice.md and my-rules.md, then paste a five-line Global Instruction ('Read my files before every task. Ask questions before executing. Show a plan first. Never delete without approval.'). Comment 'CLAUDE' for the poster's 'Ultimate Claude Guide.'

The mechanics check out. Claude Cowork is a real mode in the Claude desktop app (Mac, Windows, and since July 7, 2026 web and mobile) on Pro, Max, Team and Enterprise. Settings > Cowork > Edit Global instructions exists, and Anthropic's own Cowork course tells you to put exactly this kind of thing there: who you are, your shorthand, how you want output delivered. Cowork can be pointed at a local folder, and Cowork projects hold instructions, context folders and memory. Keeping plain markdown files about yourself and telling Claude to read them is a legitimate, documented pattern; Claude Code's CLAUDE.md is the formal version of it. What is off: the script is a near word-for-word lift of Ruben Hassid's free March 6, 2026 post (same folder names, same three files, same five-line instruction) with 'Opus 4.6' swapped for 'Opus 4.8.' Opus 4.8 exists (May 28, 2026) but Anthropic lists it as legacy; Opus 5 replaced it July 24 and Fable 5.1 sits above that since September 1. On the current models the switch is labelled 'Thinking' under Effort, and on Opus 5 and Fable 5.1 it cannot be turned off, so 'turn Extended Thinking on' is a step from spring. Nothing in Anthropic's docs says Cowork auto-reads about-me.md; the 'Read my files before every task' line is what does that work, and it is context, not enforcement. 'Sounds like you every time' is oversold; Anthropic's own docs say instruction files are followed, not guaranteed. The funnel: no Menler website, store or guide could be found anywhere public beyond the menler.in card shown in the reel. 'Comment CLAUDE' is a DM-capture mechanic; the likeliest payload is the free Hassid material, which itself gates downloads behind a Substack signup. Honest route: do the setup from Anthropic's free course, skip the comment.

What holds up

  • support.claude.com 'Get started with Claude Cowork' (Sep 2026): Cowork is a mode selected in the message box on paid plans (Pro, Max, Team, Enterprise); desktop on macOS/Windows; Global instructions live at Settings > Cowork > Edit.
  • academy.claude.com 'Standing context: Global instructions and projects' (Anthropic course): recommends putting 'who you are and what you do; the shorthand and acronyms you use... how you like output delivered' in Global instructions; projects carry Instructions, Scheduled tasks, Context folders/links, and Memory.
  • platform.claude.com Models overview (Sep 2026): current lineup is Fable 5.1 (Sep 1, 2026), Opus 5 (Jul 24, 2026), Sonnet 5, Haiku 4.5; Opus 4.8 sits under 'Legacy models (still available)' with its page reading 'Legacy. Released May 28, 2026... consider migrating to Claude Opus 5.'
  • anthropic.com 'Introducing Claude Opus 5' (Jul 24, 2026): Opus 5 'provides greatly improved performance for the same cost as its predecessor, Opus 4.8' and is 'the new default model on Claude Max, and the strongest model on Claude Pro.'
  • support.claude.com 'Change the model, effort, and thinking settings': the toggle is 'Thinking' (or 'Extended') under Effort; 'Thinking cannot be turned off in Claude when using Claude Fable 5.1 or Claude Opus 5'; it can be toggled on Opus 4.8 and earlier.
  • Ruben Hassid, Substack note Mar 6, 2026 and LinkedIn post (3,055 reactions, 313 comments): identical script: 'Opus 4.6,' 'Turn on Extended Thinking,' 'Claude-Work' with About me/Project/Template/Outputs, about-me/my-voice/my-rules, and the exact five-line Global Instruction; downloads via how-to-ai.guide, which redirects to a Substack signup.
  • code.claude.com 'How Claude remembers your project': instruction files are 'context, not enforced configuration... there's no guarantee of strict compliance.'
  • Funnel check Sep 6, 2026: three searches and two direct fetches found no website, store or guide tied to 'Menler' beyond the menler.in profile card in the reel; the Instagram handle of that name is an unrelated inactive account.

What doesn’t

  • Opus 4.8 is named as if current; Anthropic marked it legacy on July 24, 2026, six weeks before the reel.
  • The script is copied nearly verbatim from a free, six-month-old Ruben Hassid post with one model name changed.
  • 'Comment CLAUDE' leads to a DM funnel with no findable guide behind it.
  • 'Every answer stops sounding generic every time' overstates what instruction files do; Anthropic says they are followed, not enforced.

The catch

The setup is free and worth an hour, but it only works if you actually write a real my-voice.md with samples of your own writing and keep updating the rules; a paid Claude plan is required, and the model advice in the reel is already stale.

How to actually do it

  • Open the Claude desktop app on a paid plan, choose Cowork in the message box, and pick Opus 5 (default on Max, strongest included on Pro); leave Thinking alone, it is on by default.
  • Create one working folder with About Me / Projects / Templates / Outputs and connect it in Cowork; write about-me.md (what you do day to day), my-voice.md (three pieces you wrote plus phrases you hate), my-rules.md (ask first, show a plan, never delete without approval).
  • Settings > Cowork > Edit Global instructions: paste your name, role, 'Read the About Me folder before every task,' and your output preferences; add a correction there every time you repeat one.
  • Make a Cowork project per client or recurring deliverable so instructions, context folder and memory stay scoped, per Anthropic's Cowork course at academy.claude.com.
  • Skip the comment. Take the free Anthropic 'Introduction to Claude Cowork' course; if you want the original folder template, it is Ruben Hassid's March 2026 post, free to read.

Expect noticeably less generic drafts within a week if my-voice.md contains real samples, with occasional drift you correct by adding a line to Global instructions; nobody gets 'sounds like you every time.'

Confidence
High
Posted by
Menler (AI-tips Facebook page, 'India's only Claude-first learning brand')

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.