Bug: --serverUrl CLI flag has no effect on API client requests

[codxbrexx]

Description

In metacall-deploy, the --serverUrl (-u) CLI flag is parsed but applied to the internal configuration after the api client has been initialized. As a result, the provided custom URL is completely ignored during the execution of deployment commands.

The API client falls back to the default baseURL (e.g., https://api.metacall.io or the cached config). Currently, the only way to override the target URL locally is by using the hardcoded --dev flag, which always sets the target to localhost:9000.

Steps to Reproduce

1. Execute metacall-deploy using a custom --serverUrl, for example:

   metacall-deploy --serverUrl http://custom-faas-server.local:9000 --workdir .

2. Observe the network traffic or CLI output. The CLI connects to the default API (https://api.metacall.io) instead of the specified custom URL.

Expected Behavior

The API client should instantiate using the base URL provided via the --serverUrl parameter, allowing deployments to a custom self-hosted or remote FaaS instance.

Actual Behavior

The API client instantiates using the default configuration, and the --serverUrl parameter is applied to the script's config state at the very end of index.ts, well after the API client has been utilized.

Root Cause

In deploy/src/index.ts, the api object is initialized with config.baseURL:

However, the override from the CLI arguments is only checked at the end of the script line 178:

	if (args['serverUrl']) {
		config.baseURL = args['serverUrl'];
	}
})();

Proposed Fix

Apply the configuration override before initializing the API client:

const config = await startup(args['confDir']);

// Fix: Apply --serverUrl override before API client creation
if (args['serverUrl']) {
	config.baseURL = args['serverUrl'];
}

const api: APIInterface = API(
	config.token as string,
	args['dev'] ? config.devURL : config.baseURL
);

I'm willing to resolve this issue.

Note: The --dev flag should be removed in the future.