I eventually settled on the following approach:
• Install and configure Herdr (https://t.co/OFoqS1gxzi) on the host machine.
Think of Herdr as a tmux-like session manager for AI agents, allowing agent sessions to persist independently of your client connection.
👇
This keeps agent work isolated and makes it easy to reconnect to the exact session you were using, regardless of which device you're connecting from.
When I want to inspect or edit code, I connect to the same host using VS Code's Remote SSH extension.
👇
Recently, I've been working on a better setup for using coding agents like Codex, Claude Code, and Pi across multiple devices.
My goal was to run the agents on a dedicated host machine while controlling them from my laptop, so I could work from anywhere in the house. 🧵
@rikeshbaniya@deepseek_ai@claudeai 350m tokens for 6$ crazy but deepseek is usually token hungry. Would be even better if you share what task was it used on
Trying to build my own opinionated agent harness orchestration.
->Building upon the pi agent sdk.
-> Has custom tools
-> web search/fetch capabilities
-> Design’s mock
etc….
Will be writing about it soon with a demo 🙃
@fullybearded_ To be honest it is a bit worrying, which is why I am working on a feature where you can select the chrome profile that u want to give access to and some guard rails to the operations that it can perform.
Every AI browser tool I've used is slow, burns through tokens, and spins up its own Chrome, making me log in again every single time.
So I built a browser harness for the pi coding agent package that fixes all of that, and gives the agent way better access to the page. 👇
@fullybearded_ • Drives your real Chrome with your logins — no re-auth.
• Clicks by coordinates from an a11y snapshot — works through iframes & shadow DOM.
• Parallel reads, auto-queued writes.
• Native to pi, no MCP overhead; direct HTTP is 10–50× faster than the browser.