baocin

@rauchg preperc.com - predict the perc result of a piece of land. Currently only TN while I get the data pipeline nailed down. Claude Opus 4.6, Deepseek v4 flash, and Hermes agent to auto email counties.

Show me the thing you’ve built with AI you’re most proud of. Reply with a working product URL and what model / agent you primarily used.

Security things from the last few days: - CopyFail (linux pwn'd) - CopyFail 2/Dirty Frag - 13 advisories in Next.js - Over 70 CVEs addressed in MacOS 26.5 - ~50 CVEs addressed in iOS 26.5 - YellowKey (Windows Bitlocker pwn'd entirely) - GreenPlasma (Windows privilege escalation) - CVE-2026-21510 and CVE-2026-21513 confirmed to be used by Russia for Windows RCE - CVE-2026-32202 separately confirmed to be used by Russia for sensitive document access - Mini-Shai Hulud (over 300 JS and Python packages compromised via GitHub Action cache poisoning) - Google confirms they have identified AI-powered exploitation of zero days in an unidentified "open-source, web-based system administration too" - Canvas (popular LMS used in most schools) pwn'd entirely - PAN-OS (palo alto networks) pwn'd with a 9.3 severity CVE-2026-0300 Are you scared yet?

@cognition @cerebras Sounds decent.

Intelligence at 1000 tokens per second, right in your terminal. Now available with SWE-1.6 Fast, powered by @cerebras. We're giving the first 100 people who respond a free month of Max to try it out.

@Teknium My typical pattern is to have hermes make a system cron script that only elevates to hermes via the local webhook. Bam - intelligence only when needed.

Traditional cron jobs are great for silent tasks on a machine, and Hermes Agent cronjobs are great for extending that to your agent, but why not utilize the gateway and hermes' cron to access things that don't need to cost an agent's time across any messenger service you have connected? Just `hermes update` and ask your agent to setup cronjobs that need no agent in the loop, like running a systems diagnostic script that reports info to you every 12 hours, pulls in an RSS feed, and send it over to you on telegram, whatever you can think of, while saving a lot of money on purely programmatic tasks! If the script has no output, you wont get a ping either PR: github.com/NousResearch/h…

@njbotkin The pressure is real, plus expectations are sky high due to AI hype. Reliability is secondary to shiny...

I think the reason the software devs are now feverishly vibe coding is because everything is suddenly low hanging fruit. gotta grab all that fruit before it rots

Poor little 48GB Macbook... local gepa optimization runs can be no joke.

@DaveShapi Data will be ambient.

@elonmusk This is the last time in history where humans will not take reusable launch for granted. Please allocate some time at SpaceX to safeguard documentation. Your employees are at the edge. The nuance involved in each decision will never be as understood as it is today.

@DirtyTesLa ts.la/michael87494

If you have ZERO Tesla referrals, reply to this post with your referral link If you need a referral link for $1,000 off a Cybertruck, pick one of the links below!

@dsmiley411 The typo in the header really kills it though

CodeStrap's new site is live. This started as a v0 project but GPT-5.2 actually did a damn good job with this rebrand. Logo and style guide were done by a human designer. Pretty good results for a model just working from the style guide. codestrap.com

Coding agents are great.

Sometimes the 'flatter the user' reverse prompt injection by models is actually welcome and motivating.

This is the last time in human history that software will have decent human readable documentation.

@DirtyTesLa Because it is unique training data. And cost. If the AI could adapt around the smear/dirt then great, no cleaning. Now they know the limits and the car can handle dirty lenses on the offchance it runs out of cleaning fluid.

This has been bothering me. For YEARS owners have pointed out the need for camera washers due to our real world driving. But Tesla didn't listen. They waited until they put 10 supervised robotaxis out and then went "ah shit these cameras are getting dirty." Why

@aelluswamy Sweet sweet vindication

Robotaxi rides without any safety monitors are now publicly available in Austin. Starting with a few unsupervised vehicles mixed in with the broader robotaxi fleet with safety monitors, and the ratio will increase over time.

Use it to make fun stuff like: youtube.com/watch?v=wG7tEs…

Google Maps requires internet. OpenStreetMap dumps are 130GB. So I built ptiles: the entire US in 1GB. Self-host it, keep it private, includes POI data. github.com/baocin/ptiles

@simplifyinAI And if you just need building footprints github.com/protomaps/PMTi… is easily self-hostable (literally a file in S3/B2/R2) in 130GB (for the world).

@simplifyinAI OSM Buildings actually covers the US osmbuildings.org

One line of code. That’s all it takes to get access to Google Open Buildings, the largest building dataset, for any country. 100% free and available globally.

@chesterzelaya Local models. Continuous active Context aggregation for local models. That is the short term future that will unlock truly great Agents that operate proactively.

so much non-utilized compute on iphones we need more creative solutions to make use of this wasted resource