Keeping Knowledge Free
International
Joined March 2009
- Tweets27.4K
- Following7
- Followers15.8K
- Likes578
Hack In The Box retweeted
Meshticulum Alpha 002 is out - adds an opt-in, host-backed MCU DATA ciphertext queue, durable
physical-result tracking and drain, request-keyed serial events and queue-aware retry timing github.com/l33tdawg/meshticu… #meshticulum #lora #meshcore #reticulum #alpha #opensource
Hack In The Box retweeted
First fine tuned model for #sage #lantern is done -github.com/l33tdawg/lantern-… (and wtf is SAGE Lantern? A computer that fits in your hand [sage-foundation.net/hardware…]) #sage #ai #governance #meshticulum #lora
Hack In The Box retweeted
Meshticulum 1.0.0-alpha - Two fresh 100,000-byte files completed at 1.95–2.03 kB/s (15.6–16.2 kbit/s) of useful application throughput, including sender startup. Scope-to-closure rates were 1.76–1.86 kB/s github.com/l33tdawg/meshticu…
Hack In The Box retweeted
Find the weakness. Follow the clues. Capture the flag. Take on the SINCON 2026 CTF, 15–16
Oct. Get your invite code at the SINCON registration booth.
#SINCON2026 #CTF
Hack In The Box retweeted
Been playing with @DeepSeekHarness and it's pretty good but I notice that DSFV4 made more mistakes when compared to running it within codex-cli or claude. This aims to close the loop the agent kept falling through: a diff after every edit, verification on edit, claims checked against source - seems to make things better. Try it: github.com/l33tdawg/dsh-work…
Hack In The Box retweeted
Works great :)
Get started today:
brew install incoai/tap/splash
splash serve --model incoai/Qwen3.8-27B-Splash
That is the whole setup. Point your agent at it: Claude Code, OpenCode, Codex, or Hermes.
M3 or newer, macOS 26.4+, 36 GB. Blog: inco.ai/blog/splash
ALT Two commands. That is the whole setup. brew install incoai/tap/splash, then splash serve --model incoai/Qwen3.8-27B-Splash. The terminal shows the model downloading and verifying, then Splash is ready on localhost port 8000. M3 or newer, macOS 26.4 or later, 36 GB unified memory. Point Claude Code, OpenCode, Codex, or Hermes at it, or use it in LM Studio from day 0.
Hack In The Box retweeted
This looks interesting github.com/ihubanov/hunch cc @pdp
Hack In The Box retweeted
EIDOLON: Provable Delegated Agency for Faithful Digital Twins. A governance layer that lets a person delegate a cryptographically bounded, revocable, fully-attributable slice of their authority to an AI agent that decides the way they would: eidolon.onyxcreator.com/pape…
Hack In The Box retweeted
What the heck is going on here? Any explanations for why I'm suddenly getting CN text in output @thsottiaux @OpenAI? Google says it translates to "<heartbeat> Beijing Racing Plan, Caishen Zhengba Agent" and the second is "How to play Dafa Fast Three" #codex #China #AI
Hack In The Box retweeted
Mynah version 1.5.0 should be out later this week. Give it a try; find a bug, submit a PR :) l33tdawg.github.io/mynah #ai #sage #agent #signal
Hack In The Box retweeted
OGs are in the house.
We’re proud to announce Dhillon “L33TDawg” Kannabhiran as a Keynote Speaker for ROOTCON 20.
Founder and CEO of Hack In The Box, Dhillon has played a pivotal role in shaping the cybersecurity community across the Asia-Pacific region. His passion for bringing together researchers, hackers, and defenders has left a lasting impact on the industry.
#ROOTCON20 #KeynoteSpeaker #HackInTheBox
Hack In The Box retweeted
QuietType 1.0 is out 🚀
Press Fn, speak naturally, and get polished text wherever your cursor is—AI chats, email, docs, terminals and more.
Runs locally on Apple Silicon with WhisperKit/Core ML. No cloud transcription.
l33tdawg.github.io/quiettype…
#transcription #private #local
Hack In The Box retweeted
I don't know about you, but I'm tired of typing to my agents - it's too slow; tried some cloud voice transcription services but; voice + cloud == bad idea, so I fixed it l33tdawg.github.io/quiettype
#ai #agent #transcription #nocloud #privacy #security #prompting
