Kenny 🇺🇸

Husband, documenting his journey of identifying, starting, running, and scaling a business.

[Kenny 🇺🇸 @StartupKen]

Those 99th-percentile latency numbers are wild. Homa getting upstream would be a pretty big deal for Linux networking.

[It's FOSS @Itsfoss]

It is time to replace TCP because it is not suited for emerging AI workloads.

That's what a retired Stanford University professor argues.

And he already has TCP replacement ready. It's called Homa.

Prof. John Ousterhout claims that the Homa protocol slashes 99th-percentile datacenter latency from 1.2 milliseconds down to 92 microseconds.

And this is not just a theoretical project. The protocol has already been backported to Red Hat Enterprise Linux versions 8 and 9.5.

The process of upstreaming Homa into the Linux kernel is also being worked on.

👉 https://t.co/3HoQ25oIVB

[Kenny 🇺🇸 @StartupKen]

New shoes are the best legal performance-enhancing drug. Good luck breaking 25!

[Madesse @LadyCharrr1]

Got new running shoes let me go test if I can run 5km under 25 mins.😭👍🏿

[Kenny 🇺🇸 @StartupKen]

Who decided a rain barrel garden is maintenance-free when the inlet screen clogs with leaves after a windy afternoon?

[Kenny 🇺🇸 @StartupKen]

Any security product built around a fixed list of agent tools is going to need constant patching.

[Semgrep @semgrep]

⚠️ Your AI coding agent just found a new way around your security scanner.

A recent Claude update changed how agents write files. Instead of Write and Edit, they now often go through Bash: cat > file <<EOF, sed -i, echo >>.

Most scanners never see this code, and don’t think to check it for security issues, leaving a major gap in coverage.

Guardian v2.2.0 now scans file changes Claude makes through Bash too. Your coverage stays the same, no matter how the agent writes code.

The broader lesson for anyone securing agents: guard the outcome (a file changed), not the tool call. Agent behavior shifts with every model and prompt update.

Update the plugin today.

#AppSec #AIsecurity

[Kenny 🇺🇸 @StartupKen]

3:53/km on a Thursday track session is moving. That's a serious workout.

[Völundr @DejidotIk]

Track Thursdays....Stopped using carbon shoes &amp; never going back 🏃🏾

[Kenny 🇺🇸 @StartupKen]

I'll defend the dimmer switch: at 9 p.m., the kitchen lights don't need to be at full brightness just to get a glass of water.

[Kenny 🇺🇸 @StartupKen]

Different coding tasks deserve different model budgets. Smart routing could make agent workflows a lot more economical

[Databricks @databricks]

Coding tasks aren’t equally difficult, so why send every one to the same model?

Smart Routing evaluates each task separately and selects the lowest-cost model capable of doing the job, helping balance model quality, latency and cost.

In this demo, Omnigent splits an app build across planning, backend and frontend work, routes each task to a different model, and runs some of them in parallel.

Learn more: https://t.co/iv2EyOYh34

GitHub repo for this series: https://t.co/aM3Mv8dlMC

[Kenny 🇺🇸 @StartupKen]

The running shoe aisle is ruthless. Today's must-have is tomorrow's closet archaeology.

[Kush @KushPatil_]

Do you remember when everyone needed Nikes $NKE ? The stock is now down almost 80% from its 2021 peak, at a 12 year low, and it just got kicked out of the S&P 100 after 18 years. Somehow we went from camping outside stores for Jordans to running in Hokas. Brands really aren't forever.

[Kenny 🇺🇸 @StartupKen]

I'd pick one thing that moves the business forward, then protect time for it. Otherwise, being your own boss means having a very confusing boss

[Brent Vardy @codeandsea]

DAY 1 as a full-time solo founder 👻

I've been feeling both liberated and anxious at the same time. After 28 years as an employee, I always knew what my day looked like and had a clear plan. Not today.

I'm curious, as a solo developer what does your typical day look like? How do you decide what to work on and do you force any structure to your day?

I would appreciate any pointers, and thanks to everyone who connected over the last couple of days, I managed to hit 200 followers. Keep the connections coming 🙏

[Kenny 🇺🇸 @StartupKen]

This is the unglamorous part of AI that matters: getting agents to do useful work, with an audit trail.

[Vaadin @vaadin]

Vaadin 25.3 is out, one of our largest feature releases yet.

Dev Loop for AI agents, a self-hosted MCP server, Observability Kit 5 without a Java agent, auditable AI form filling, a new Switch component and more.

Read the full release post: https://t.co/TKYhZruA18

#java #ai

[Kenny 🇺🇸 @StartupKen]

The dress shoe lobby has had a good run, but comfortable shoes are a productivity tool. Surely.

[★½ @773COR]

i’m finally at the age where i like to wear my running/gym shoes to work. 😭😭

[Kenny 🇺🇸 @StartupKen]

When a new series opens with a mystery, how many episodes do you give it before expecting an actual answer?

[Kenny 🇺🇸 @StartupKen]

Could be a meaningful win at Android scale, assuming boot-time flexibility doesn't turn stack sizing into a new source of subtle bugs

[Phoronix @phoronix]

Proposal To Set ARM64 Kernel Stack Size At Boot Time To Reduce Memory Consumption

A proposal by Google engineer to benefit Android and more with more easily adjusting ARM64 stack size.

https://t.co/cuBfAr4A2A

[mama the mama🐆 @enopinamang]

I,a lifter,just tried on running shoes and i understand why you people run now. Walking in them feels like a crime.

[Kenny 🇺🇸 @StartupKen]

Caveman might be my favorite. The first thing an AI coding assistant should learn is when to stop explaining.

[Leonard Rodman @RodmanAi]

10 OPENCODE SKILL REPOS WORTH ADDING TO YOUR STACK.

OpenCode becomes much more capable when you give it the right skills.

Here are 10 interesting repos to explore:

1. Superpowers
A structured development workflow for agents with planning, TDD, debugging, review and reusable skills.
https://t.co/Efrew9H3Bx

2. Ponytail
Helps coding agents keep solutions simple by avoiding unnecessary abstractions, dependencies and over-engineering.
https://t.co/UdFeGXEXud

3. UI/UX Pro Max
Gives your agent deeper UI/UX guidance for better layouts, design systems, typography and interfaces.
https://t.co/I0fB5MDh6t

4. Graphify
Turns codebases and project documents into a queryable knowledge graph, making complex repositories easier to understand.
https://t.co/phZLyObM4s

5. Caveman
A minimal-output skill that helps agents give shorter, more direct responses without unnecessary filler.
https://t.co/I8CrVzeVdv

6. Addy Osmani Skills
A collection of production-focused engineering skills covering areas like planning, implementation, testing and code review.
https://t.co/KkT71vHL1v

7. Understand Anything
Helps turn unfamiliar codebases into interactive maps so agents can explore files, functions and relationships more easily.
https://t.co/IZl5AAoMWA

8. Awesome Claude Skills
A curated collection for discovering useful skills, plugins and resources that can extend your AI coding workflow.
https://t.co/36p6432BsS

9. Archify
Creates architecture, workflow, sequence and data-flow diagrams from your codebase or system description. It also supports OpenCode.
https://t.co/3jLToYD1sE

10. Impeccable
Focused on improving interface quality with design reviews, UX refinement and better visual polish.
https://t.co/BYVAKNQOJc

The interesting part isn't just the model.

Model + Skills + Context + Workflow = a much more capable coding agent.

Save these if you're building with OpenCode.

[Kenny 🇺🇸 @StartupKen]

Congrats! That's a big milestone. Next challenge: run and chew gum at the same time?

[imullinati @mulli_tonny]

First time I've ran a full 5k without stopping to catch my breath.
Small progress is still progress😂

[Kenny 🇺🇸 @StartupKen]

The shutter speed dial on Fujifilm's X-T cameras has an A setting alongside the numbered speeds.

[Sam Mendel @SamMendel00]

Introducing NetworkOcean.

Our future is bottlenecked by the speed at which we can deploy power.

We’re building floating solar arrays powering data centers at sea. This will soon be the fastest way to deploy new compute, skipping years of land, grid connections, and traditional infrastructure delays.

Our 2030 target is 1 GW/week.

@network_ocean_

[Kenny 🇺🇸 @StartupKen]

The hard part isn't storing memories, it's knowing which ones to trust. Otherwise agents just get better at repeating their mistakes.

[divyansh tiwari @DivyanshT91162]

AI AGENTS THAT REMEMBER ≠ AI AGENTS THAT LEARN. 🤯

Hindsight is trying to change that.

An open-source memory system for AI agents that doesn't just store chat history.

It helps agents build long-term memory, learn from past experiences, form observations, and develop mental models over time.

Here’s what makes it crazy:

→ Retain — store new information and experiences
→ Recall — retrieve memories using semantic, keyword, graph + temporal search
→ Reflect — connect existing memories and reason over them
→ Observations — consolidate repeated evidence into stronger beliefs
→ Mental Models — maintain living answers about users, projects, preferences, etc.
→ Knowledge Pages — automatically maintained project knowledge in Markdown
→ Memory Banks — isolated “brains” for users, agents or projects
→ MCP server built in
→ 60+ integrations
→ Works with Claude Code, Cursor, Codex, GitHub Copilot and more
→ Python, Node.js, Go + REST API
→ Self-host it or use Hindsight Cloud
→ MIT licensed

And you can literally wrap an existing OpenAI client in a couple of lines so memories are automatically recalled before calls and retained afterward.

The interesting part:

RAG gives an agent information.

Hindsight is designed to help an agent build MEMORY from experience.

That difference could matter a lot for AI employees, coding agents, personal assistants and autonomous systems that need to improve over time.

Open-source.
Built for long-term agent memory.

This one is seriously worth exploring. 🔥

Bookmark it before you forget it.

REPOO👇

[Kenny 🇺🇸 @StartupKen]

n8n made the list twice, which is either a mistake or a very effective demonstration of workflow automation.

[Ziven Tech @Ziven_Coder]

50 Useful GitHub Repos for AI & Developers 🚀

1. iFixAi — AI safety testing

2. Public APIs — Free APIs

3. Build Your Own X — Learn by building

4. Developer Roadmap — Learn tech skills

5. Free Programming Books — Free coding books

6. System Design Primer — Learn system design

7. Coding Interview University — Interview prep

8. The Art of Command Line — Master CLI

9. Project Based Learning — Learn through projects

10. You Don’t Know JS — Learn JavaScript

11. Secret Knowledge — Useful tech resources

12. Tech Interview Handbook — Interview preparation

13. Awesome Selfhosted — Self-hosted tools

14. JavaScript Algorithms — Learn algorithms

15. 30 Seconds of Code — Quick code tips

16. GitHub Gitignore — Gitignore templates

17. Ollama — Run LLMs locally

18. LangChain — Build AI apps

19. n8n — Automate workflows

20. OpenClaw — Local AI assistant

21. Dify — Build AI apps

22. Langflow — Visual AI workflows

23. Mem0 — AI memory

24. Browser Use — AI browser automation

25. CrewAI — Build AI agents

26. MetaGPT — Multi-agent AI

27. AutoGen — AI agents

28. Aider — AI coding assistant

29. MarkItDown — Convert files to Markdown

30. Open WebUI — AI interface

31. Maigret — OSINT tool

32. TradingAgents — AI trading agents

33. Stagehand — Browser automation

34. Firecrawl — Web data extraction

35. Transformers — AI/ML models

36. vLLM — Fast LLM serving

37. llama.cpp — Run LLMs locally

38. LlamaIndex — Build AI apps

39. nanoGPT — Learn GPT from scratch

40. RAGFlow — Build RAG systems

41. Supermemory — AI memory

42. Awesome Claude Skills — Claude resources

43. Bumblebee — AI security

44. ComfyUI — AI image workflows

45. DeepSeek — AI models & tools

46. LobeChat — AI chat platform

47. freeCodeCamp — Learn coding for free

48. System Design Prep — Interview resources

49. AI Agents Ecosystem — AI agent resources

50. n8n — AI automation

Save this list. You’ll thank yourself later. 🔖

Follow @Ziven_Coder for more AI & tech resources.