Founder/builder.
Not ex-OpenAI, not ex-Anthropic, not ex-Google.
Just tired and shipping. Building DealsForge, CanRunIt & Mozart . AI tools, local model, op.
Hermes Agent said: "the agent that grows with you."
I loved that.
So I asked:
what if it was one binary,
local-first,
with model routing,
real verifier review,
git checkpoints,
and replayable runs?
Meet Sparrow 🐦
The slightly chaotic pirate cockpit for coding agents!
Local first.
Cloud when it makes sense.
Everything replayable.
Full demo ↓
Who else is tired of choosing between local and cloud?
Launch in 04:00:00 🚀
Introducing NVIDIA Nemotron 3 Ultra.
A frontier smart open model built for long-running agents that need to plan, reason, use tools and keep working across complex coding, research and enterprise workflows.
Up to 5x faster inference and up to 30% lower cost for agentic tasks.
Learn more: https://t.co/h9XLqqYPFf
AI coding is powerful.
But shipping AI-generated code without a security pass is becoming one of the fastest ways to create production risk.
Build fast!
But don’t ship blind!
I audited 12 repos generated with Cursor/Copilot this week.
11 had critical security issues.
Here are the patterns I see everywhere.
And how to fix them in 2 minutes 👇
The common pattern?
Developers using AI to code often don’t think in security terms.
LLMs don’t think in security terms either.
Result:
Code that works locally… and explodes in production...
Has anyone received a quick response for the codex Pro program for open-source projects?
I applied a while ago, and now is exactly when I need these capabilities, because the project is launching tomorrow.
Thanks in advance to the whole community for helping boost this message so OpenAI can reply.
You’re the best!
You can now train 120B+ parameter models locally on a laptop! 🔥
We collabed with NVIDIA and Microsoft to bring LLM training on the 128GB unified memory RTX Spark laptop!
Inspired by the best ideas from
✅️ Claude Code
✅️ Codex
✅️ OpenCode
✅️ OpenClaw
✅️ Hermes Agent.
But the goal is narrower:
trust, routing, rollback, replay, and local/cloud freedom ❤️ !
Hermes Agent said: "the agent that grows with you."
I loved that.
So I asked:
what if it was one binary,
local-first,
with model routing,
real verifier review,
git checkpoints,
and replayable runs?
Meet Sparrow 🐦
The slightly chaotic pirate cockpit for coding agents!
Local first.
Cloud when it makes sense.
Everything replayable.
Full demo ↓
Who else is tired of choosing between local and cloud?
Launch in 04:00:00 🚀
I’ve been cooking something truly unique for a while.
A new contribution to the community.
Something that will help you build differently.
Faster. Stronger.
Countdown started — deployment in 6:00:00
@Teknium if you don’t mind sharing with us and saving me from reading the full update note because it’s very late here 😅
A few questions:
1. What search method is used under the hood?
2. Does efficiency depend on the model selected for the agent?
3. Is there any safeguard against false negatives?
4. Can we keep a small critical set of tools always loaded?
Thanks in advance for the answers!
You are the GOAT you think you are 🐐😎
Now in Hermes Agent, you can load an arbitrary amount of tools from mcps to plugins and get access to it all, with no compromise to your context!
Anytime you load more than 10% of your context window in tools from mcps or plugins, this feature will activate automatically!
hermes update to get access now 😎