Kevin here, indie dev (lorisLabs.)

Thorttle started as a menu bar meter for Claude code usage, and grow into a cockpit !

you have the full claude meter as expected with no telemetry, no network. and that in the free version.

Pro is where it got interesting :

Cockpit have every claude code project on your Mac in one window. Each embedded terminal using SwiftTerm, auto hibernate to keep a 16gb Mac from getting out of memory.

Remote Control that hand a session off to any linux server (I use proxmox) I have a IOS and Vision Pro app coming up soon!

my main goal and what I prefer: Claude code burn a lot of token you dont even see... bloated claude.md files & stale memory, dead skills, barbs output ...

Throttle optimiser with a AI assistant that audit your claude.md and settings.json hook with a real read_file tool call emit search/replace...

autopilot applied automatically concise output style (65/75% fewer output tokens!!!) usage statusline, auto -archive of orphaned project memory, riskier one (stale memory cleanup, dead skill pruning) stay opt in and guarded.

session start + [re compact hook that prune irrelevant memory before it every reacts the context window.

All this is measurable not guessed! you have a token saver and stats, mine reads 78M saved this week again 1b used !

EVERYTHING RUN LOCAL or in your own iCloud private database. No server of mine ever see your transcripts or usage or telemetry... and if you want just the free meter : https://github.com/lorislabapp/throttle-meter (MIT)

for all this saving stuff and pro version features, 29€ one time 2 Mac 30days no questions refund ! (not joking it saving me a lot !)

I WOULD LOVE FEEDBACKS!!! Which optimisation would you trust an app to do automatically and with must stay a reviewed patch? The usage math comes from local JSONL, not claude.ai's API does it match what you see?? 29€ one time for people paying 100/200USD a bond for max, fair? -- Anything else really

Thank for looking!