We currently only support a subset of our command grammar for our custom parsed actions. We would like to:

• Set up our test harness to try running all spoken forms from our recorded tests through our custom command parser to check that the output payload matches the command payload
• Expand the parsed command grammar so that we pass all of the above tests

Note that we need to think about whether we want to support all spoken forms. The following examples seem awkward:

• Decorated marks, eg "take air". In this case, using the spoken forms for characters seems a bit strange. In a written grammar, it seems more natural to do something like "take a"
• Ordinals / numbers, eg "take second token". Might be fine for smaller ones, but it's a bit strange for larger ordinals / numbers to spell them out
• Plurals, eg "take two tokens"

Keep in mind also the following relevant issues to our parser:

• Create spoken form widget for use in docs #2232
• Automatically generate human-readable description of command #2096
• various applications discussed in Expand cursorless talon command api #492