One place to talk to all your AI agents.
Agent Hub is one place to talk to all your AI Agents. Claude Code, Codex, Hermes, OpenClaw, running locally or across every VPS you own, in one warm little native app.
Bring the AI agents you already use.
Claude Code, Codex, Hermes and OpenClaw, plus any OpenAI-compatible endpoint. Each runs locally on your Mac or over SSH on a remote box.
Claude Code
Codex
Hermes
OpenClaw
Terminal
Pick an AI agent. Try it yourself.
Type in the composer, hit / for slash commands, click Approve or Deny on tool calls.
The full AI-agent toolkit, minus the tab-switching.
Everything the agent CLIs give you on the terminal: Claude Code, Codex, Hermes, OpenClaw. Packaged into one calm native macOS UI.
Local or SSH. Same chat window.
Point Agent Hub at omar@prod-1.fly.dev and it runs the CLI there. Your box in NYC, your GPU in Frankfurt, your laptop on the couch, all as threads in the sidebar.
Granular permissions.
Ask per-tool, auto-approve safe edits, or bypass entirely. Your rules, per agent.
Slash commands.
Fuzzy-find every command registered by every agent, keyboard-first.
Paste images, drop files.
Screenshots from the clipboard, files from Finder. They go to the agent, not into the void.
Hot-swap models.
Pick the right model per thread. Codex effort levels, Claude variants, Hermes sizes, all live.
Three commands to install Agent Hub.
Clone the repo, install, launch. Or download the macOS .dmg if you'd rather skip npm.
Agent Hub FAQ, before you clone.
Does it work on Windows or Linux?
Which agents are supported?
Do you send my data anywhere?
How do permissions work?
Can I use my own models or a custom endpoint?
Is it really free?
How is Agent Hub different from Cursor or Continue.dev?
Where does Agent Hub store API keys?
Can I run Claude Code on a VPS from Agent Hub?
Chat with every AI agent from one macOS window.
Free, open-source, MIT. Clone Agent Hub from GitHub, build from source, or grab the .dmg.