How to pass parameters via Advanced option for the Nextflow config

I am trying to run the Oxford Nanopore epi2me/wf-basecalling pipeline using Seqera Platform and AWS Batch.
The nextflow.config includes base.config, which specifies these parameters when using AWS Batch:

    awsbatch {
        process {
            executor = 'awsbatch'
            queue = "${params.aws_queue}"
            memory = "16 GB" // likely not enough!
            withLabel:wf_common {
                container = "${params.aws_image_prefix}-wf-common:${params.wf.common_sha}"
            }
            shell = ['/bin/bash', '-euo', 'pipefail']
            
            // lift limit on simultaneous gpu jobs for cloud
            // and ensure that the host mounts relevant driver bobbins inside the container
            withLabel:gpu {
                maxForks = null
                containerOptions = "-e NVIDIA_DRIVER_CAPABILITIES=compute,utility --gpus all"
            }
            withLabel:wf_basecalling {
                container = "${params.aws_image_prefix}-dorado:${params.wf.container_sha_basecalling}"
            }
            withLabel:wf_bonito {
                container = "${params.aws_image_prefix}-bonito:${params.wf.bonito_sha}"
            }
        }
    }

The parameter aws_image_prefix is set to ‘null’ in nextflow.config

So within Seqera Platform, I specified this advanced option for the Nextflow config file:

params {
           aws_image_prefix  = 'ontresearch'
}

However, this does not appear to work. The resolved configuration in Seqera Platform shows:

aws_image_prefix = '[secret]'

And the pipeline fails with

Error executing process > 'prepare_reference:cram_cache (1)'

Caused by:
  Invalid container image name: [:]-wf-common:shabadd33adae761be6f2d59c6ecfb44b19cf472cfc

which suggests that the aws_image_prefix parameter isn’t being properly set.

How do I properly pass parameters to override default values in nextflow.config?