Sr Agentic Engineer, Full Stack Dev, DevOps, suffering from LLM psychosis .
Joined July 2024
- Tweets7.3K
- Following1.5K
- Followers1.5K
- Likes20.7K
Bessi S retweeted
THE SWITCH 2 HAS BEEN JAILBROKEN BY AI 🚨
pepsipu just released ns2jb, the first known jailbreak for the Switch 2. It's software-only and works on the latest firmware, 23.0.1.
The craziest part is that Codex did most of the hacking. He hooked both Switches up so the AI could send controller inputs, watch the screen, and even press the power button with an Amazon smart light switch whenever the console crashed. His prompt was literally "set a goal to jailbreak the switch 2. make no mistakes".
The way in was a bug in SmileBASIC, a Switch 1 game, which it then chained into Nintendo's own file system service.
Readers added context they thought people might want to know
The developer clarified the quoted prompt was a "silly caption" not actually used, and the jailbreak required substantial human effort over ~2 weeks (AI assisted with some bug-finding and the demo mod).
pepsipu.com/blog/2026-10-s…
gitlab.com/pepsipu/ns2jb
gbatemp.net/news/switch-2-…
openai's updated usage policies, effective nov 20, require model routers and AI gateways to get openai's review and approval before offering direct access to its models. curious how many of the routers sitting under coding agents make the cut openai.com/index/broad-acces…
Bessi S retweeted
API version 2026-10-21 is now available.
The Claude API has evolved a lot, but this is our first official version release since 2023. We plan to release versions more often so you can handle breaking changes before a new model requires them.
2023-06-01 is not deprecated.
Omnara is an open-source managed agent platform, a self-hostable alternative to Claude Managed Agents and OpenAI's Agents API. Agents run on any model and machine with tools, secrets and permissions built in, launched from a YAML config. 3,063 stars.
github.com/omnara-ai/omnara
Bessi S retweeted
OpenAI models are coming to Google Cloud.
We're bringing our frontier models and Codex agentic workflows to Google Cloud's Model Garden, so enterprises can use our technology where they already build.
openai.com/index/openai-mode…
Bessi S retweeted
October has been an absolute nightmare for software
> Railway - customer secrets exposure
> LiteLLM - malicious PyPI + Docker releases
> Zeabur - infrastructure breached
> OpenCode - AI API keys exposed
> OVHcloud CLI - official binaries backdoored
> Rank Math PRO - malware in official plugin
> TeamUpdraft - customer data compromised
> ConnectWise - cloud backup identities compromised
> Tensorlake - npm package infected
> SubQuery - backdoored npm package
> GhostAction - GitHub repos with malicious workflows
> PHP - critical RCE vulnerabilities
> Node.js - security vulnerabilities patched
> eMudhra - certificate authority keys compromised
> Tailscale - eMudhra incident
At this point, the biggest security risk might not be the code we write, but the code we trust...
someone trained a 360M model to be a terminal. no shell or filesystem behind it, it just predicts the next screen from your keypresses. it runs vim and tmux and usually remembers your files a few commands later. type pythoon and it gives itself away.
simedw.com/2026/10/21/world-…
github says agent traffic set the pace for moving git to azure. 99.99999% of repos now live in both places; the busiest are held back by a 25ms round trip on every serialized write. funny to see coding agents turn into a data center power problem.
github.blog/engineering/arch…
PandaCode is a native desktop app that runs Claude Code, Codex, OpenCode and other coding CLIs side by side, with Git, a terminal and a browser in the same window. An architect agent can hand tasks to up to three workers. 408 stars.
github.com/aicodeking/pandac…
PHP2Shell: after an out-of-memory error, PHP leaves corrupted state that the next request reuses. decade-old bugs, and Wiz says frontier models got exploits running on WordPress, Drupal and Joomla in hours. "too fiddly to exploit" was carrying a lot wiz.io/blog/php2shell-php-rc…
Decks lets you edit slides by commenting on them: leave a note under a slide and Claude Code or Codex changes the deck, then replies on the slide with what it did or a question. A Checks tab runs five- and ten-second tests and title tests. 96 stars.
github.com/paulmil11/decks
litellm 1.107.0 on PyPI shipped a cook.pth that runs a credential stealer whenever Python starts, even without importing litellm. the sdist's only visible change was one maturin line, data = "cook", and pins matched rc2, so a dependency diff looks clean safedep.io/litellm-pypi-comp…
Bessi S retweeted
If you are wondering whether your website is affected, we have packaged a Codex skill to help check.
Point your Codex at this repo to guide you through it: github.com/openai/php2shell-…
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordinated…
Readers added context they thought people might want to know
OpenAI may claim a role in the coordinated response and patching process. But the initial discovery and exploit development, however, occurred several months earlier by an independent researcher.
x.com/lcfr_hax/statu…
Bessi S retweeted
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordinated…
Readers added context they thought people might want to know
OpenAI may claim a role in the coordinated response and patching process. But the initial discovery and exploit development, however, occurred several months earlier by an independent researcher.
x.com/lcfr_hax/statu…
Bessi S retweeted
PHP 8.4.27 released - 89 fixes in 30 components, including security fixes (CVE-2026-109618, CVE-2026-109619).
Changelog + downloads: php-net.pro/news/php-8-4-27-…
OpenAI models and Codex are coming to Google Cloud's Model Garden, and customers can pay out of their existing GCP commitments. that drawdown is how this gets past procurement. odd to see them headed into Google's own Antigravity too. openai.com/index/openai-mode…
agentplay replays your coding agent sessions as a map and a timeline: what it read, what it edited, which commands failed, where the context filled up. Runs locally with uvx and works with Claude Code, Codex, Cursor and more. 278 stars.
github.com/Entelligence-AI/a…
Bessi S retweeted
Today we’re announcing a new collaboration with @nvidia to bring AI to industrial engineering:
- faster simulation
- open benchmarks and supporting datasets for physics AI
- evaluations for agentic AI in CAE and EDA
More here: mistral.ai/news/mistral-nvid…
