Oh my, it wired up its own autopilot and is wrecking havoc now.
3,502 karma · joined September 29, 2017
Github: /kordless and /deepbluedynamics (agent orchestration and terminals)
Reach me at:
kord @ deepbluedynamics
or
kord @ gmail
Build on!
Oh my, it wired up its own autopilot and is wrecking havoc now.
This has been covered before: https://aisle.com/blog/ai-cybersecurity-after-mythos-the-jag... (https://news.ycombinator.com/item?id=47732020)
> Anthropic’s cautious roll-out was justified. The problem with publicly releasing models, however, is that guardrails can be jailbroken, and apparently that is exactly what happened shortly after the release
The future is unevenly distributed. Anthropic, and Amodie in particular, seem to be of the mind they can control a bit of the unknown using words. They are likely being guided by the very product they built. *AI CAN MAKE MISTAKES
That Project Glasswing bullshit reeks of it. Corporations have take control of our attention, our Internet, and now our thinking.
I say it's high time to take it back.
However, banning that type of content is an arms race waiting to happen. It also neglects the fact that powerful software can be built with AI with proper attention and knowledge of the domain.
It's an attention problem: https://deepbluedynamics.com/blog/rise-of-the-systemmer
Rust-based hybrid search engine: https://github.com/DeepBlueDynamics/lume and https://github.com/DeepBlueDynamics/shivvr
Agentic terminals: https://github.com/DeepBlueDynamics/hyperia
And, he has a working solution.
It's 2026, FWIW.
I have a friend who is a graphic designer/market strategy guy and he's been using Anthropic to build sites and even did an agent on his own page that helps guide the user through onboarding. I reviewed the code a few times and gave him some tips and it looks pretty good and works flawlessly.
He maintain a lot of customer's sites (design wise) and all the customers are responsible for their own hosting and ssl certs. He got tired of them calling him about expired ones, so he had Claude write a script and use Agentmail to notify him when one expires.
A few of them were needing updating when he wrote it, and when I reviewed it (with Claude Fable) it discovered that in the event they were all up-to-date, it wouldn't email him. Other than that, it works perfectly and runs on his machine on a schedule.
This morning he had it write a script to monitor his computer for load, after having issues with Adobe.
Cloudflare provides services that any of their users may enable to charge crawlers for access. And they announced a while ago that this was going to be enabled by default on new domains, whether the customer asked for it or not (and there is plenty online about people wondering why they aren't being crawled). When it is enabled, Cloudflare either blocks, require registration of the crawler (forcing enabled billing), or provide one themselves: https://developers.cloudflare.com/changelog/post/2026-03-10-...
My compressed claim was close enough (in my opinion), given the substantive argument was broader: chokepoint, MITM, decryption, the IP vs AI tension.
It's the same pattern as Flock cameras quietly becoming the default license plate surveillance layer across US cities. Private infrastructure is becoming the substrate for something that used to be public or distributed, with the concentration happening faster than any deliberative process about whether we want it.
I'm willing to be wrong on any specific piece of this. But chasing down receipts for every line is effort, and I'm tired and old. The shape is what matters, and the shape is a man-in-the-middle.
Almost everything I build will run locally or on Google Cloud in serverless mode.
The README on the repo for the app is has been updated with instructions for you. You will need Docker.
We're a country of growing "ever blamers".
Something is wrong? Blame someone else. Distract. Deflect. Deny.
Find a head to roll. Find a scapegoat. Use it as personal leverage to gain power.
Fuck the ever blamers.
The fact is, Cloudflare is a man-in-the-middle. That's their focus, that's their purpose.
They will limit your local crawler from accessing pages. They will demand you use their crawler.
They will decrypt your traffic if they get a warrant. They always decrypt your traffic anyway, but they will give it to state actors if they demand it.
That's not to say anyone should break the laws, but the issue right now is that intellectual property is incompatible with what is coming with AI.
I don't hate on Cloudflare because it's a bad service. It's actually pretty good, but the fundamental problem is they make their purpose to be a single choke point of all data on the Web.
That's not right. It never was.
I built an entire local-first sovereign agentic stack on Linux that completely replaces the IDE-centric model with a terminal control plane called Hyperia. Hyperia is a terminal emulator with a decoupled agent sidecar that hooks directly into standard protocols like the Model Context Protocol.
Instead of just reading passive text buffers, it monitors discrete command lifecycle events across your shell sessions and web panes, catching stack traces as your test suite runs like a true pair-programming partner. To make this safe and reproducible, you cannot let an LLM run arbitrary, side-effect-heavy tool calls directly on your host machine.
This is handled by Nemesis, a container orchestration runtime that acts as a secure, session-persistent sandbox for the agentic workspace. When an agent writes code or touches system files, it executes inside an isolated Docker container, keeping the host operating system completely pristine.
For data-heavy tasks like parsing local markdown files or indexing an entire photo archive, you should avoid proprietary cloud vector databases. This stack uses Shivvr, a local semantic search engine that handles chunking and inverted vector embeddings entirely on your own hardware so your data never leaves your laptop.
Finally, the extraction, ingestion, and scraping of local docs or web sources is handled by Grub, an automated, high-speed crawler that feeds structured data back into the system. Modern tooling shouldn't mean chaining yourself to a proprietary cloud SaaS platform.
By exposing standard Unix primitives like files, shells, and local compilers to an API, sandboxing the environment in a container, and letting a local agent orchestrate the workspace, the cloud-vendor magic fades away and actual sovereign software engineering takes over.
All of this is a WIP but I use it every day to work on it.
I do provide cloud support for somethings like embeddings and crawling, but you can run it local if you want. The only thing closed source is the memory system, but it still runs local if you want it.
FWIW, I'm working on Nemesis8: https://github.com/DeepBlueDynamics/nemesis8 if you want to team up. I'm kordless at gmail or kord at deepbluedynamics
It’s a bunch of people complaining about things they don’t actually control. The only logical move is to change perspective and go build something.
Yes, I will do pretty much whatever it takes to call attention to what I built. You call it “slop,” but I suspect you never bothered to look. Instead, you took the vibe of the thread — that people often post their own thing out of context — and applied it to someone making counterarguments.
In this case, I’ve built a lot of open-source software. A fair amount of it has been worked on by me for over five years. That includes real agent infrastructure: Hyperia, a local/self-owned agent control plane; N.U.T.S. services for crawling, embeddings/search, auth, tunnels, and backend glue; and a broader stack aimed at sovereign AI systems people can actually run, inspect, and control themselves.
That’s not slop.
It’s real code. It does real things. It is actually open source.
This place is supposed to be for people who build and hack. Not people who sit around blaming, sneering, and mistaking cynicism for taste. Trying to steer things back toward building and hacking is everyone’s job here. Being insulting is not.
Do better. Build better. Drop the fake empathy for things online.
Just adding a menu this morning to allow re-attaching and session management.