AgentX is a terminal chat agent built on the OpenAI Responses API over WebSocket transport.
For most users, the simplest path is: install it globally, run agentx-setup once, then start agentx.
It also includes an experimental web GUI, but that is still a proof of concept and a lot of things are broken or incomplete.
A separate agentx-setup helper can edit .env values and manage the Linux GUI service.
If you installed the package globally, use:
agentxIf you are working from the repository root, use:
node agentx.mjsThis is mainly for repository-local development, not the global install path:
npm run start:guiThen open the local port it prints, usually http://localhost:3100.
Use the CLI if you want the more reliable path; the GUI is still rough.
agentx-setupUse that to edit .env values and manage the GUI service on Linux.
agentx --helporagentx -hshows quick helpagentx --versionoragentx -vprints the package versionagentx --debugprints raw websocket logs and suppresses live status lines
Set your API key in the shell environment, or let agentx-setup write it into a local .env file for you:
export agentx_api_key="your-key-here"
# or: export AGENTX_API_KEY="your-key-here"AgentX prefers agentx_api_key and falls back to AGENTX_API_KEY.
The launchers also load .env when present.
- Start AgentX.
- Type a normal message and press Enter.
- AgentX sends that message to OpenAI.
- The response streams into the terminal as it is generated.
- The prompt shows your current working directory.
- AgentX waits for your first message before contacting OpenAI.
- Tool calls may stream command arguments and shell summaries live.
- If
.agentx_responseidexists, the session resumes automatically. If it contains pending tool calls, AgentX asks how you want to continue.