igorhvr
Joined 18 years ago
Karma
698
Submissions
114
Member since
Oct 5
Recent submissions
Comment · 14 days ago
I think matchlock ( https://github.com/jingkaihe/matchlock ) is the right design for this context (at least for secrets and keys) - a combination of virtual machines (not just containers, it is built on top of firecracker) with credential swapping just like iron proxy - the agents do not have access to the secrets at all but they receive placeholders instead and their network calls are intercepted and secrets are injected when the network requests happen... still does not solve the exfiltration problem you mentioned, though.
Comment · 5 months ago
I use PI ( https://pi.dev ) and ( https://hermes-agent.nousresearch.com/ ) as the main drivers together with deepseek-v4-pro as the main model (~10M/day tokens overall there).
Hermes basically rules my personal life at this point - it is a _very_ useful personal assistant.
I also use it at work (integrated at Slack) and at this point it answers most of both my emails and slack messages (I calibrated https://github.com/blader/humanizer with a large corpus of my own voice to make it less annoying for the others). My routine now involving walking in circles while exchanging messages with hermes directing it how to answer this or that... Hermes uses an llm-wiki ( https://gist.github.com/karpathy/442a6bf555914893e9891c11519... ) as a source of information when drafting suggested replies - I have a cronjob that feeds it all emails, slack messages, meeting minutes every single day.
Claude Code with Opus 4.6 for multimodal/vision, design and writing tasks ("Create a crisp memo from this meeting transcription" is a prompt that will bring great results with either Opus 4.6 or GLM-5.1) - very recently I started to use https://github.com/anomalyco/opencode occasionally with opus models too (I am forcing myself a little bit because it is hard to help people with a tool you are unfamiliar with).
For building software automatically I currently use one of the harness above for for launching https://tamandua-tetradactyla.nfshost.com/ feature-dev-merge-worktree runs (it provides workflows on top of PI+deepseek).
Where it comes from: until recently I used https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16d... for automatic software building but while planning an AI bootcamp I concluded that teaching Gas Town along with everything else would be impossible (too hard/complex), and decided to teach https://github.com/snarktank/antfarm instead but did not want to add OpenClaw as one more dependency - so I built Tamandua ( https://github.com/igorhvr/tamandua ) and I ended up using it all the time! I now every single day before going to sleep launch a couple of runs and it is very cool waking up to see them done.
For autoresearch-like, optimization, and other tasks with a very clear measurable goal (such as increasing test coverage, changing things from one programming language to another, etc) I use https://github.com/davebcn87/pi-autoresearch (100% of the time on top of deepseek-v4-pro).
For debugging or very hard problems I use codex w/ GPT5.5. I don´t like its personality (lazy) but I do think it is smartest model available. As evidence, here is a commit of a problem where I tried Opus 4.8, Deepseek-v4-pro and a couple of other models and they all failed to understand what the bug was: https://github.com/NousResearch/hermes-agent/pull/38198/chan... - once the bug was found within codex I launched from it a tamandua bug-fix-merge-worktree run on top of deepseek-v4-pro that created the commit itself...
As a web application I use OpenWebUI and I am specially fond of the notes feature ( https://docs.openwebui.com/features/notes/ ) which I did not find anywhere else.
Last but not least, I love playing with local models. https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct is my current favorite for coding and deepseek-r1 for general tasks. I also started yesterday testing https://github.com/antirez/ds4 - it works _very_ well from what I could see so far.
What comes next? Trying to figure out what is the "deepseek-v4-pro of multimodal" model (frontier performance, efficient/comparatively cheap to run, support for image/audio/video/etc). Currently using kimi-k2.6, will test Minimax M3 soon.
Ah, almost forgot: https://huggingface.co/microsoft/VibeVoice-ASR will give you AMAZINGLY good meeting transcriptions (my hermes vibecoded a program to use it). Seriously, night and day difference from what the big players provide natively in their platforms. Have 8 people talking in 3 different languages? No problem - you will need a bit of patience and beefy hardware, only..
Comment · 7 months ago
On the contrary, they could have stated their points much more bluntly and strongly than they did in the post. I had the same impression upon reading it.
Milquetoast perfectly describes it, I am happy to see less common words used around here (specially when the convey the intended meaning this precisely), and I find claiming "affectation" of the person who used it unnecessarily rude.
Comment · 1 year ago
This is so cool - thank you! I have a very (ahem) useful purpose for this: I use a command line application that calls back to a browser during authentication and that alone prevented me from doing what I needed/wanted from an ssh terminal... I will now happily laugh my ass off as it launches firefox from inside my terminal every time I use it.
Comment · 1 year ago
This one is customizable in Lisp, and overall pretty neat: https://sawfish.tuxfamily.org/ - I have been happily using it daily for many years now. :-)
Comment · 3 years ago
Comment · 4 years ago
Thanks! Yggdrasil ( https://yggdrasil-network.github.io/ ) should probably in this list too, except that it doesn't need a central beacon node.
Comment · 4 years ago
If you don't mind having to add the solar kit for power (separate purchase) https://www.accairwater.com/home-atmospheric-drinking-water-... and similar products could be used..
Comment · 5 years ago
Ha - so cool. I love that it can be easily used to send a stream (the only thing I missed in magic-wormhole). Basically,
ls -lah | croc send
Thanks for this link!
Comment · 5 years ago
AFAIK folks are adding this as a joke. Example: https://web.archive.org/web/20210705124241/https://jobs.leve...
Comment · 5 years ago
I thought the new title better but I did not want to change the original (because I had already shared the link with friends), so I just changed when submitting. Should've gone with the original...
Comment · 5 years ago
The thing is they change/deprecate/retire paid-for services too (in a brutal contrast with the competiton).
Two quotes from one of the posts referenced in the submission (from Steve Yegge):
...I know I haven’t gone into a lot of specific details about GCP’s deprecations. I can tell you that virtually everything I’ve used, from networking (legacy to VPC) to storage (Cloud SQL v1 to v2) to Firebase (now Firestore with a totally different API) to App Engine (don’t even get me started) to Cloud Endpoints to… I dunno, everything, has forced me to rewrite it all after at most 2–3 years, and they never automate it for you, and often there is no documented migration path at all. It’s just crickets. And every time, I look over at AWS, and I ask myself what the fuck I’m still doing on GCP. ...
... Update 3, Aug 31 2020: A Google engineer in Cloud Marketplace who happens to be an old friend of mine contacted me to find out why C2D didn’t work, and we eventually figured out that it was because I had committed the sin of creating my network a few years ago, and C2D was failing for legacy networks due to a missing subnet parameter in their templates. I guess my advice to prospective GCP users is to make sure you know a lot of people at Google… ...
Comment · 5 years ago
About deprecation: the gold standard is what AWS is doing with SimpleDB: essentially, never.
The thing here is that if you run hundreds of services in production - many of which work smoothly and you don't need to touch often, you will find that Google's habit of forcing you to change how to use their tooling will generate a huge burden...
They discourage you from using it and make it clear that for every use case some other tool at AWS would be best, and have been doing so for several years now... They wont even list it anymore under https://aws.amazon.com/products/databases/ ..
Still, they support it ( https://aws.amazon.com/simpledb/ ) because there are customers with legacy systems that depend upon this service.
Comment · 5 years ago
Submitter (& author) here. I just changed the submission title to exactly match the original.
Do you love Scheme?
Checkout out this library: http://github.com/igorhvr/bedlam