ninjæon

Technologist/Biohacker

[ninjæon @ninjaeon]

@PawelHuryn I love this and have been following it since the 1st release in Oct. Any chance it could be updated for muse-wonder-1.0-contributor and the $200 muse plan? Thank you!

[ninjæon @ninjaeon]

@AIatMeta Love that this was done with publically available models and the research papers are shared. Open source the scaffold please? 😘

[ninjæon @ninjaeon]

@claireszhou @Answeror Update: Restriction seems to have been lifted! Haven't seen any communication, but everything seems to be working again! THANK YOU MUSE CODE TEAM!! 🙏🤎 @claireszhou @Answeror

[ninjæon @ninjaeon]

@claireszhou @Answeror my Muse Code account was restricted today ("repeated policy violations") with no email or reason given. Appeal filed and /feedback sent (mbfb_d4amdxnj7wh4jqb4x5tsoydwyq). Could someone take a look? Happy to share details by DM. Thanks!

[ninjæon @ninjaeon]

@axhiii88547 GWBASIC.EXE

[ninjæon @ninjaeon]

@claireszhou Thank you!! Was down to about ~17% for the week after giving up ~60% allowing non-contributor max thinking subagents run all night on Power 🤎

Here's a vote for increasing the thinking effort limit on the contributor model from xhigh to Max (w/o decreasing the rate limit) 🙏

[ninjæon @ninjaeon]

@borgerntendies @claireszhou Pt.2: glm-tray optional - poller writes meta-usage-quota.json directly; mod can read that (skip JSONL relay). taskbar-ai-quota is modified (adds Meta, 401-fatal exit, container restart detection). browserless CDP wrappers spawn temp profiles — login dies w/restart.

[ninjæon @ninjaeon]

@borgerntendies @claireszhou Camoufox Docker (persistent profile, :9380 API + :6080 VNC for one-time login). Poller (pythonw @5min) → tab API evaluate() runs in-page fetch of subscription-quota → JSON → glm-tray → Windhawk taskbar-ai-quota. TLS gate: only real browsers. Profile volume = login survives.

[ninjæon @ninjaeon]

@PrathyushPrasad didn't they decrease the quota burn when using the contributor model this week? people probably slamming it afterwards. plus the published rate limit is 100 requests or 3,000,000 tokens per minute on it, which is kinda easy to hit if you are running full context windows

[ninjæon @ninjaeon]

@wiederitzschLE @pablobh @myuill @closte_cloud It's now back up normally for me in Southeastern USA.

Earlier I couldn't get the VPN method to work, but the hosts method did. If copying and pasting to hosts, remove https from the pasted text

[ninjæon @ninjaeon]

@pablobh @myuill @closte_cloud Once you have admin access to your Closte website, if you goto Litespeed Cache Settings > Cache > Enable Cache > OFF and save, it should fix any issues with the images loading on the website.

[ninjæon @ninjaeon]

@BT Love that you're still rockin' the ethereal BX5a's!