Testing

cli.test() gives you a programmatic harness for the real CLI. It runs the same code paths, but replaces stdin, stdout, stderr, and process exit behavior so your tests stay deterministic.

Vitest example

tests/cli.test.ts

import { describe, expect, it } from "vitest";
import { createCLI } from "@oscli-dev/oscli";

describe("init flow", () => {
  it("stores the provided project name", async () => {
    const cli = createCLI((b) => ({
      description: "test",
      prompts: {
        name: b.text().label("Name"),
      },
    }));

    cli.command("init", async () => {
      await cli.prompt.name();
      cli.success(`Created ${cli.storage.name}`);
    });

    const result = await cli.test({
      argv: ["init"],
      inputs: { name: "my-app" },
    });

    expect(result.storage.name).toBe("my-app");
    expect(result.output).toContain("Created my-app");
    expect(result.exitCode).toBe(0);
  });
});

TestOptions

Prop

Type

TestResult

Prop

Type

Test flow

Use this order when writing tests around a CLI.

Create the CLI in the test

Build the same CLI object you use in production code.

Run `cli.test()`

Pass inputs, flags, and argv to drive the flow.

Assert on the result

Check storage, flags, output, and exitCode directly.

Test mode always behaves like non-TTY output. Prompt summary lines are still written to the captured buffer so you can assert on them.