ncode

ncode is a fast, simple coding agent for your terminal.

It talks to any OpenAI-compatible endpoint, runs as one Rust binary, and writes inline to normal terminal scrollback instead of redrawing a fullscreen UI.

Simplicity and speed are the whole point.

Install

Download the binary for your machine from Releases, then:

chmod +x ncode
./ncode

On first run, ncode creates:

~/.ncode/config.toml

Put your defaults there:

[model]
model = "gpt-4o"
base_url = "https://example.com/v1"
# api_key = "sk-..."

You can also pass config per run:

ncode --model gpt-4o --base-url https://api.openai.com/v1 --api-key sk-...

Commands

• /model picks from the endpoint's model list.
• /thinking shows or hides reasoning.
• /new starts fresh.
• /resume opens an old session.
• /help shows the rest.

Project defaults can live in ./ncode.toml. They override ~/.ncode/config.toml.