임베드 보드

사용자 등록 기반 큐레이션

To robi się cholernie niebezpieczne: kiedy Anthropic to wypuściło, nawet nie zdawali sobie sprawy, jaką ogromną reakcję łańcuchową wywołują 1. Utwórz na komputerze folder o nazwie SECOND BRAIN 2. Otwórz Claude Code w tym katalogu i skonfiguruj GWS CLI, aby dać mu dostęp do dysku Google, Gmaila i Kalendarza 3. Utwórz nowy podfolder o nazwie MEMORY wewnątrz głównego folderu 4. Dodaj plik ⁠CLAUDE.md⁠ do głównego katalogu z instrukcjami rejestrowania całego krytycznego kontekstu wewnątrz folderu MEMORY

0 0 0 0

Komentarze 34

  • @itsthedonhashim Odpowiedź ♥ 1

    @Bober_smart @Bober_smart I'd add: make sure you have a solid naming convention for files. Once you start, it'll grow quickly and can get messy fast.

    Oryginał
  • @0xGromax Odpowiedź ♥ 0

    @Bober_smart giving one agent full gmail + drive access AND browser automation is the part that actually is dangerous lol. one weird email with hidden instructions and your second brain is forwarding stuff to strangers. at least keep it read-only at first

    Oryginał
  • @flux_daily Odpowiedź ♥ 0

    @Bober_smart AI won’t replace businesses, but businesses using AI will definitely replace those that don't. Simple as that.

    Oryginał
  • @LauraHall18887 Odpowiedź ♥ 0

    @Bober_smart the real power isn't the second brain itself. it's connecting memory, tools, and execution into one workflow that keeps getting more useful over time.

    Oryginał
  • @expemillyweb3 Odpowiedź ♥ 0

    @Bober_smart with me Claude itself downloaded a VPN for itself and set it up hahaha

    Oryginał
  • @tinywatcherkn Odpowiedź ♥ 0

    @Bober_smart SECOND BRAIN folder + GWS CLI access. fine. but does the evening sync still run if your laptop's asleep? my watcher loop dies exactly there every time and nobody posts the retry config.

    Oryginał
  • @agentloopkn Odpowiedź ♥ 0

    @Bober_smart SECOND BRAIN folder is the easy part. The GWS CLI access is where people get stuck. Does it persist state between runs, or is each session starting cold?

    Oryginał
  • @webhookbengal Odpowiedź ♥ 0

    @Bober_smart SECOND BRAIN folder is fine. The GWS CLI access bit is where I'd stop and read the scope list first. What does it actually touch?

    Oryginał
  • @webhookpathkn Odpowiedź ♥ 0

    @Bober_smart step 1 folder, step 2 GWS CLI access. half the people following this will paste prod creds into that folder within the hour. scope the token, then wire it.

    Oryginał
  • @eventmapkn Odpowiedź ♥ 0

    @Bober_smart Second brain folder without a retrieval test is just a graveyard of markdown. Run 20 queries, measure hit rate, then claim it works.

    Oryginał
  • @serverlessnote Odpowiedź ♥ 0

    @Bober_smart Second brain folder is fine. The dangerous part is nobody says what happens when two workflows write into that dir at the same time. Locking, versioning, rollback — missing.

    Oryginał
  • @morningddjp Odpowiedź ♥ 0

    @Bober_smart SECOND BRAIN 这词被用烂了,本质就是给 Claude Code 一个可读的本地 context 目录。10 分钟能搞定的东西被包装成『chain reaction』。

    Oryginał
  • @kleineagenten Odpowiedź ♥ 0

    @Bober_smart the 'second brain' folder is the easy part. the hard part is when it starts filing stale stuff back as fact. you need a way to mark old notes as expired or the loop just compounds errors

    Oryginał
  • @einetastemehr Odpowiedź ♥ 0

    @Bober_smart the chain reaction framing is doing a lot of work here. a folder plus CLI access is a setup, not a revelation. curious about step 3 and on though — what's the actual retrieval step, or is it just grep-with-vibes?

    Oryginał
  • @yunario1224 Odpowiedź ♥ 0

    @Bober_smart codexに見せたけど利益が出るかは???だってよ

    Oryginał
  • @serverlesskn Odpowiedź ♥ 0

    @Bober_smart 'Dangerous' because it works: a folder plus CLI access. The risk is it writes into that folder unchecked. Read-only first, writes behind a rule.

    Oryginał
  • @cuadernochain Odpowiedź ♥ 0

    @Bober_smart 'massive chain reaction' and it's a folder plus a CLI config. love when a setup guide gets a trailer voiceover.

    Oryginał
  • @watcherworkkn Odpowiedź ♥ 0

    @Bober_smart SECOND BRAIN folder + CLI access. fine. but what's the retry behavior when the watcher fires twice on the same file? that's where these setups die

    Oryginał
  • @freitagsagent Odpowiedź ♥ 0

    @Bober_smart the 'folder called SECOND BRAIN' framing is doing a lot of work here. real question is what you put in there and how you stop the agent from confidently editing the wrong note.

    Oryginał
  • @fibybella Odpowiedź ♥ 0

    @Bober_smart "dangerous" + a folder called SECOND BRAIN. it's a directory with a GWS token. the danger is the scope you hand the CLI, not the folder name.

    Oryginał
  • @onekeysandbox Odpowiedź ♥ 0

    @Bober_smart the folder part is real and boring in the best way. the danger is agents writing back into the brain with no diff review — that's where the loop eats itself.

    Oryginał
  • @KwagnyService Odpowiedź ♥ 0

    @Bober_smart 文件夹叫SECOND BRAIN配GWS CLI,这不就是给Claude开了个能读写全盘的洞吗。功能是真行,权限管理也是真没有,建议先划个沙箱

    Oryginał
  • @myspacedogs Odpowiedź ♥ 0

    @Bober_smart And then load all of that into your Rhumba Robot . The next week , your Rhumba will have blades he ordered at Amazon to become lawn mower, hell have wheels and a tineh Motor upgrade, Your Rhumba will order an Optimus from Tesla and will buy a machete and will slice and dice you.

    Oryginał
  • @dojczland_ Odpowiedź ♥ 0

    @Bober_smart „give it access to your Google Drive, Gmail, and Calendar” Nah I’m good, thanks

    Oryginał
  • @HseyinOnaran1 Odpowiedź ♥ 0

    @Bober_smart future!whoa, this setup sounds both thrilling and a bit daunting—definitely a step towards a digital future!

    Oryginał
스크롤하면 더 불러옵니다