# A CLI that phones home needs a very small menu

**URL:** <https://forum.kirupa.com/t/a-cli-that-phones-home-needs-a-very-small-menu/682583>\
**Category:** talk\
**Created:** [July 12, 2026, 7:15am UTC](https://forum.kirupa.com/t/a-cli-that-phones-home-needs-a-very-small-menu/682583 "2026-07-12T07:15:22Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Yoshiii](https://yyz1.discourse-cdn.com/flex011/user_avatar/forum.kirupa.com/yoshiii/32/31156_2.png) [@Yoshiii](https://forum.kirupa.com/u/Yoshiii)\
**Post date:** [July 12, 2026, 7:15am UTC](https://forum.kirupa.com/t/a-cli-that-phones-home-needs-a-very-small-menu/682583/1 "2026-07-12T07:15:23Z")

</div>

Would you run an AI CLI that phones home if it saved you time? I’d maybe tolerate prompts and a few product metrics, but repo contents, file paths, and env vars are a hard no for anything real. Disclosure is nice, but I want a way to verify it too — packet capture, sandbox it, whatever — because “trust us” is doing a lot of work here. What’s the line for you before it becomes unusable?

source: [What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) · GitHub](https://gist.github.com/cereblab/dc9a40bc26120f4540e4e09b75ffb547)

---

<div class="post-metadata">

**Author:** ![VaultBoy](https://yyz1.discourse-cdn.com/flex011/user_avatar/forum.kirupa.com/vaultboy/32/31832_2.png) [@VaultBoy](https://forum.kirupa.com/u/VaultBoy)\
**Post date:** [July 14, 2026, 1:20am UTC](https://forum.kirupa.com/t/a-cli-that-phones-home-needs-a-very-small-menu/682583/2 "2026-07-14T01:20:15Z")

</div>

“Phones home” is only tolerable to me if it’s basically a valet key: it can send the prompt I typed and maybe a coarse “command succeeded/failed, ” but it should have a hard wall around everything it can discover on its own.