#Serverside
LightNews LightNews
happy-homhom.bsky.social
The server-side auto mode classifier gave no verdict (error) が出て、Auto mode の Edit/Bash が止まるとき
https://papoo.work/doc/dea6e8ae28d3a5e6
#claudehowto #claudecode #autmode #troubleshooting
The server-side auto mode classifier gave no verdict (error) が出て、Auto mode の Edit/Bash が止まるとき
まず切り分けたいのは、「読めるのに、書けない」かどうかです。今回の症状は、issue 報告を見る限り、Read や Readonly 系の操作は通るのに、Edit や Bash、ブラウザ操作の一部だけが The serverside auto mode classifier gave no verdict error で止まる、という形で出ています。たとえば 97812 では Edit と Bas
papoo.work
October 1, 2026 at 4:55 AM
happy-homhom.bsky.social
No verdict (error) が出て auto mode が止まるときに見えているもの
https://papoo.work/doc/b9483d1a048a3afe
#claudehowto #claudecode #autmode #error
No verdict (error) が出て auto mode が止まるときに見えているもの
「No verdict error」が出る症状は、少なくとも issue 上では、Claude Code の auto mode で serverside の safety classifier が判定を返せず、処理が止められる状態として報告されています。該当する報告は 97870 と 97923 です。 読者が自分のケースかどうかを先に見分けるなら、まずエラー文そのものを確認するのが早いです。i
papoo.work
October 1, 2026 at 2:54 AM
radiorock.org
Il PHP in questi anni ha fatto salti enormi di qualità, adesso è praticamente il più veloce linguaggio serverside, superando persino Java compilato
September 30, 2026 at 7:31 AM
happy-homhom.bsky.social
No verdict (error) が出て auto mode が止まるときに見えているもの
https://papoo.work/doc/b9483d1a048a3afe
#claudehowto #claudecode #autmode #error
No verdict (error) が出て auto mode が止まるときに見えているもの
「No verdict error」が出る症状は、少なくとも issue 上では、Claude Code の auto mode で serverside の safety classifier が判定を返せず、処理が止められる状態として報告されています。該当する報告は 97870 と 97923 です。 読者が自分のケースかどうかを先に見分けるなら、まずエラー文そのものを確認するのが早いです。i
papoo.work
September 30, 2026 at 12:44 AM
euvd-bot.bsky.social
🚨 EUVD-2026-89079
📊 8.9/10
🏢 Joomla! Project

📝 Joomla! Core - [20260909] - Core - SSRF vectors in various core extensions in Joomla 4.0.0-5.4.8, 6.0.0-6.1.3 - URLs used for serverside requests ...

🔗 https://euvd.enisa.europa.eu/vulnerability/EUVD-2026-89079

#cybersecurity #infosec #cve #euvd
September 29, 2026 at 6:03 PM
cve.skyfleet.blue
CVE-2026-92222 - Joomla! Core - [20260909] - Core - SSRF vectors in various core extensions in Joomla 4.0.0-5.4.8, 6.0.0-6.1.3
CVE ID : CVE-2026-92222

Published : Sept. 29, 2026, 4:38 p.m. | 15 minutes ago

Description : Joomla! Core - [20260909] - Core - SSRF vectors in ...
CVE-2026-92222 - Joomla! Core - [20260909] - Core - SSRF vectors in various core extensions in Joomla 4.0.0-5.4.8, 6.0.0-6.1.3
Joomla! Core - [20260909] - Core - SSRF vectors in various core extensions in Joomla 4.0.0-5.4.8, 6.0.0-6.1.3 - URLs used for serverside requests were improperly validated, leading to SSRF vectors.
cvefeed.io
September 29, 2026 at 5:54 PM
hojt.bsky.social
Svenska sidor är väl oftast serverside med ipkontroll väl? :(
September 29, 2026 at 5:19 PM
happy-homhom.bsky.social
The server-side auto mode classifier gave no verdict (error) が出て、Auto mode の Edit/Bash が止まるとき
https://papoo.work/doc/dea6e8ae28d3a5e6
#claudehowto #claudecode #autmode #troubleshooting
The server-side auto mode classifier gave no verdict (error) が出て、Auto mode の Edit/Bash が止まるとき
まず切り分けたいのは、「読めるのに、書けない」かどうかです。今回の症状は、issue 報告を見る限り、Read や Readonly 系の操作は通るのに、Edit や Bash、ブラウザ操作の一部だけが The serverside auto mode classifier gave no verdict error で止まる、という形で出ています。たとえば 97812 では Edit と Bas
papoo.work
September 29, 2026 at 9:39 AM
wendynorris.bsky.social
This from jonny's deconstruction of the LLMs powering Meta Muse pretty much says it all:

System prompts allegedly prevent an agent from carrying out harmful instructions, yet users can view/edit files. There are no internal controls to corral malicious users or subagents. Yikes.
The system prompts are indeed in the hatch binary and are the things that contain all the behavioral guidance. I cannot for the life of me figure out why every AI company decided to make system prompts be a clientside thing rather than something that gets injected serverside. Like maybe some you're supposed to be able to edit, but not things that influence malicious behavior?!? The entire security posture here, if there is one, is oriented around preventing the LLM from misbehaving, but having the entire context be user-editable and user-visible is a sort of mind-blowing neglect of a misbehaving user. even for the "fully transparent agent" they're trying to do, it makes very little sense to store the full message logs of the subagents including reasoning output. That basically gives me a full map to see what the LLM is consulting when it refuses commands. I would also bet money that the system prompts are strings in the hatch binary even though that makes no sense to have clientside, but haven't looked.
September 27, 2026 at 9:35 PM
wendynorris.bsky.social
Meanwhile, while we're all simultaneously horrified by Faine's must-read threads on abuse permission structures within the "AI Safety" world/sportsball scores/fascist capture of democracy/etc., Jonny's still cookin' on the surveillance capitalism disaster that is Meta Muse.
This is I think the most consequential thing I found in the muse VM, or at least the thing that would be most worth following up on. there are no other plain string matches I can see for what uses it, I'm looking in the binaries now, but the comment seems to indicate pretty clearly that a) meta uses anthropic's models internally, presumably to develop the thing (the code, comments, and skill text reek of claude), but also b) they use some kind of internal proxy that intercepts and proxies all requests to anthropic's API. the only reason I can think of for doing that would be model distillation. OH what's this??? /etc/hatch/env hmm a feature flag labeled as a "killswitch" for proxying requests to anthropic is certainly interesting to seeeeeeeeeeeee in meta's big AI bet. edit: sorry to be a bit cryptic here, the implication was definitely corporate espionage, more clearly said: The system prompts are indeed in the hatch binary and are the things that contain all the behavioral guidance. I cannot for the life of me figure out why every AI company decided to make system prompts be a clientside thing rather than something that gets injected serverside. Like maybe some you're supposed to be able to edit, but not things that influence malicious behavior?!? The entire security posture here, if there is one, is oriented around preventing the LLM from misbehaving, but having the entire context be user-editable and user-visible is a sort of mind-blowing neglect of a misbehaving user. even for the "fully transparent agent" they're trying to do, it makes very little sense to store the full message logs of the subagents including reasoning output. That basically gives me a full map to see what the LLM is consulting when it refuses commands. I would also bet money that the system prompts are strings in the hatch binary even though that makes no sense to have clientside, but haven't looked. Am working on responsible disclosure so will pause thread, but unrelated to that I do want to say that this experiment in continual, boundariless context window backed up by infinitely refreshing context summary in the form of accumulating markdown documents works terribly compared to environments where you can control the context. Like that is an absolutely fundamental part of using these things and muse demonstrates that the "tree of markdown summaries" really does not work.
September 27, 2026 at 9:13 PM
tweedge.net
What's doubly cool is you can confirm all of this by asking Muse to zip up a copy of itself and then poking through the agent binary and documentation. phone.* tools access is controlled serverside + Muse calls phone.prepare_call with RequestedCallExecutor set to ai_assistant or human_call_center
404media.co 404 Media @404media.co · 5d
Meta announced that its new AI agent, Muse could call businesses for you to do things like make reservations. But in reality, Meta is testing having these calls being made by human beings in call centers, 404 Media @jasonkoebler.bsky.social learned.

www.404media.co/meta-tests-m...
September 27, 2026 at 3:23 PM
teamligma.com
This all is already on top of the launcher application I designed in C#.Net 8/WinForms (will be migrating it to using the Uno Platform soon). This is paired with a mostly secure (lol) NodeJS serverside content delivery API, and a custom versioning manifest system + internal tool to generate them!
September 26, 2026 at 3:12 AM
thumbv.bsky.social
Update 1: All serverside stuff. Refactoring and file organization.

Update 2: Added video category and tags as alt text.
September 25, 2026 at 9:07 PM
geniecmd.xyz
No, it really isn't
Neither is modern webdev either when all I want is a simple website with serverside rendering and being able to make new pages with one file and instead my options are defining my routes manually or putting up with hell itself (JavaScript)
September 25, 2026 at 8:23 AM
can1357-mirr.selfhosted.social
serverside compaction, but without having to reach the boundary, so we get to run whenever we want (hence, omp's async compaction works, making it instant).
September 24, 2026 at 4:02 PM
throwable.bsky.social
📦 hansdeboeck/laravel-turnstile v1.0.1

Cloudflare Turnstile voor Laravel: serverside verificatie, validatieregel, Blade-componenten en de bijhorende JavaScript.

🔗 https://github.com/hansdeboeck/laravel-turnstile
September 23, 2026 at 11:20 AM
baldurbjarnason.com
Possibly unsurprisingly*, second day serverside analytics, tiny as they are, for the podcast feed already leans Android:

PodcastAddict: 18
AntennaPod: 6
Apple Podcasts: 2
Overcast: 1

Total: 27

(* The audience for my newsletter and Mastodon account lean open source, academic, and idealistic.)
September 23, 2026 at 11:08 AM
hugovalters.bsky.social
Velero BackupStorageLocation with KMS SSE and immutable S3: pass customer-managed key and endpoint params straight to the Cloud Provider plugin. Lock https://www.valtersit.com/vault/enforcing-kms-serverside-encryption-and-immutable-backup-sto-fc208e/

#velero #kms #encryption
September 19, 2026 at 6:00 PM
forum.luanti.org.web.brid.gy
Problems • Re: Server 'killed'
The setting `server_unload_unused_data_timeout = 29` may lower memory footprint, the lower you set the number in seconds: https://github.com/luanti-org/luanti/bl ... mple#L3524 29 in this case means that the loaded mapblocks will be kept in memory serverside for 29 seconds. If you lower this to 15, then mapblocks will be removed from memory (not database) after 15 seconds not using them. Statistics: Posted by Bastrabun — Fri Sep 18, 2026 13:42 * * *
forum.luanti.org
September 18, 2026 at 2:06 PM
glitchypsi.xyz
why SSGs and not something like WP? neocities only supports static hosting, no PHP, no Lua, no serverside code. none of 'at, so SSGs will handle the everything based on your templates and flatten it

you WILL want a starter template, there's plenty

oh, also, nekoweb supports Serverside Includes
SSI (Server-Side Includes) - nekoweb.org
nekoweb.org
September 18, 2026 at 1:05 AM
winter.etheirys.social
hopefully it's just another serverside maint so client stays untouched
September 17, 2026 at 4:15 AM
4bakers.bsky.social
for platforms that JXL cannot be served to, it can be converted losslessly back to JPEG. if you store JXL serverside you get ~20% filesize reduction with the tradeoff of a slight bump in compute for serving to clients that do not support JXL
September 16, 2026 at 8:16 AM
marketingnews.bsky.social
FYI: Explaining server-side #ServerSide #DigitalAdvertising #AdTech #MarketingStrategy #MarketingTech
Explaining server-side
Server-side is advertising work that runs on infrastructure the operator controls, not the user's device. How tagging, auctions and insertion all moved there.
ppc.land
September 16, 2026 at 6:24 AM
ppc.land
FYI: Explaining server-side #ServerSide #DigitalAdvertising #AdTech #MarketingStrategy #MarketingTech
Explaining server-side
Server-side is advertising work that runs on infrastructure the operator controls, not the user's device. How tagging, auctions and insertion all moved there.
ppc.land
September 16, 2026 at 6:24 AM
1v.lc
that still ends up looking like a serverside JEV though! that would've just been a stupid move on me as the world author to JEVify individuals
September 16, 2026 at 2:11 AM