AI in your terminal
Run any NeonCodex task from the command line. Stream output in real time. Pipe results into your workflow.
Install & authenticate
Install globally via npm, then authenticate with your API key from Settings → Developer.
Run your first task
Output streams in real time — just like the web interface.
Commands reference
neoncodex run "<prompt>"--model claude-sonnet-4-6 --no-streamneoncodex chat--model autoneoncodex file <path>--prompt "Summarize this"neoncodex browse <url>--analyze "Find all prices"neoncodex tasks--limit 20 --status completedneoncodex get <task-id>--outputneoncodex stream <task-id>neoncodex login--key nck_live_...neoncodex logoutneoncodex config--model auto --format markdownneoncodex versionPipe into your workflow
Use stdin/stdout to compose NeonCodex with other Unix tools.
IDE integrations
Install the NeonCodex AI extension for inline chat, right-click code actions, Fix/Explain/Security scan, and apply-diff. Or use CLI in the integrated terminal (Ctrl+`).
Download .vsix extensionWorks in IntelliJ, PyCharm, WebStorm, GoLand via Alt+F12 terminal. Pipe selected code: cat file.py | neoncodex run. Native plugin on roadmap.
Native support for bash, zsh, fish, and PowerShell. Full SSE streaming. Works on macOS, Linux, and Windows (WSL). Install once, works everywhere.
Use NeonCodex CLI in Cursor's integrated terminal alongside Cursor's built-in AI. Best of both worlds: Cursor edits + NeonCodex extended tasks.
Configuration file
Stored at ~/.neoncodex/config.json or per-project at .neoncodex.json.
{
"apiKey": "nck_live_xxxxxxxxxxxxxxxxxxxx",
"model": "auto",
"outputFormat": "markdown",
"stream": true,
"baseUrl": "https://neoncodex.io/api",
"timeout": 180
}