
lots of ideas, just enough agents to keep up software developer @aerolab https://teapawt.net/t.co/CXWjCIIAcr
Joined October 2024
- Tweets31
- Following87
- Followers5
- Likes624
T3 code is for people that are coming from native apps from their frontier labs. your capabilities are enhanced, your subscriptions are in one place and all is well.
but for true agent multiplexing and orchestration you need to be using @herdrdev - you get herdr panes that are summoned by your orchestrator (imagine a fable 5.2 orchestrator working with sonnet/ codex 5.6 workers) you also get better organization with workspaces, tab models and status icons for your agents as well + you can add features as you go on your interface + you can open a port to any of your devices (your phone/ computer)
that said T3 code is amazing if you're coming from native apps but as you start AI-maxxing you'll soon realize it isn't enough.
contur retweeted
Claude now supports more of the MCP Apps extension, such as app tools and embedded sites, so your full app can run inside Claude.
Start building: claude.com/docs/connectors/b…
Replying to @claudeai
Also new: with interactive connectors, you can now open the full version of a tool inside Claude. Work in Adobe, Atlassian, and Box beside your conversation. Make changes yourself, or ask Claude to make them with you.
Rolling out on web and desktop.
contur retweeted
Replying to @wesbos @luisrudge
Correct. I highly suspect in less than 12 months we’ll see Rex screenshots that mostly have no terminals at all.
contur retweeted
A few weeks ago I rewrote an entire Node app in Go. 28mb binary, almost instantaneous builds + deployment.
If we aren't manually typing code, what is the point of developer-friendly languages? Why not write everything in Go, or even C++?
herdr-core was built for this.
tell your main agent to read hc - - skill and orchestrate all other agents and terminals for you!
I've changed how I work over the last few weeks.
I now have one lead agent for each area of focus, running all the time, and a chief of staff agent that oversees them. Each lead manages its own sub-agents that write code, test it, chase reviews, and coordinate with team members.
Everything I need to decide or review comes into one prioritized queue, a simple local web app. Each product review includes screenshots and a video demo of the feature. I clear the queue in batches, mostly by voice and keyboard shortcuts. The leads also bring back ideas I hadn't considered.
Working from one queue, instead of switching between dozens of chats, has cut a lot of mental overhead. The system keeps learning my preferences and distilling my judgment into a "Brian brain," so its recommendations and first drafts get better before they reach me. The goal is fewer corrections per change.
Every change still goes through the same human review and security checks. It is eliminating tasks, not human judgment.
What an exciting time to push the frontier of productivity, especially as we build the economy for agents (agentic finance). We should all keep pushing in this direction.
contur retweeted
Introducing MyGo, a pure Go framework for building desktop app in webview or native UI, it's like GPUI and Tauri had a baby, but made in Go so it compiles fast
mygo.egoist.dev
contur retweeted
models have no sense of time. they'll trigger a 30 min rebuild for a 2 line change instead of thinking ahead and batching the work
current frontier RL rewards token efficiency
future RL will reward time efficiency
the fast takeoff will feel absurd after this












