Currently, all files are packaged as tar.gz or zip archives within an image layer. To retrieve action information - such as type (e.g., node, docker), inputs, outputs, ... the only method is to download the entire layer, extract it, and parse the action.yml file. This process is slow and inefficient because its involves unnecessary steps and waste of resources.
To address this, I propose storing the action.yml content within the OCI image config section, using the mediaType=application/vnd.github.actions.manifest.v1+yaml. This approach would enable development tools to easily discover action metadata without the need to download the entire image, improving efficiency and resource utilization.
Currently, all files are packaged as tar.gz or zip archives within an image layer. To retrieve action information - such as type (e.g., node, docker), inputs, outputs, ... the only method is to download the entire layer, extract it, and parse the action.yml file. This process is slow and inefficient because its involves unnecessary steps and waste of resources.
To address this, I propose storing the
action.ymlcontent within the OCI imageconfigsection, using themediaType=application/vnd.github.actions.manifest.v1+yaml. This approach would enable development tools to easily discover action metadata without the need to download the entire image, improving efficiency and resource utilization.