yeah awsx is pretty opinionated which can be fine sometimes, but if you want to do it yourself you're better off just using the base aws and docker resources directly

and i only have a c# example to give, but here is how you can create an AWS ECR repo and obtain credentials to pass to the pulumi.Docker.Image resource:

var ecrRepo = new Pulumi.Aws.Ecr.Repository(...)

var ecrCreds = ecrRepo.RegistryId.Apply(registryId => Pulumi.Aws.Ecr.GetCredentials.InvokeAsync(
    new PulumiAws.Ecr.GetCredentialsArgs
    {
        RegistryId = registryId,
    },
    invokeOptions));

var registry = ecrCreds.Apply(creds =>
{
    var decodedCredentials = Encoding.UTF8.GetString(Convert.FromBase64String(creds.AuthorizationToken)).Split(':');
    return new Pulumi.Docker.ImageRegistry
    {
        Server = creds.ProxyEndpoint,
        Username = decodedCredentials[0],
        Password = decodedCredentials[1],
    };
});

var dockerImage = new Pulumi.Docker.Image("image", new Pulumi.Docker.ImageArgs
{
    ImageName = Output.Format($"{ecrRepo.RepositoryUrl}:{your-tag-or-build-id}"),
    Registry = registry,
    Build = dockerContext (path to dockerfile or other options)
});

then you can stick

dockerImage.BaseImageName

in your ECS task definition

re-reading your original question, it looks like you are creating the ECR Repo in a different process - so you might just need to modify what i have above to take in the ECR registry id / arn as an input or something and start from there