MVP CLI (step in event loop) to go from set of instrument/measure requirements to state in system card

[uittenbroekrobbert]

The check would be a function with a signature like this:

def func(set of requirements (instrument urns), state system_card):
    return diff_of_urns_of_tasks_that_need_to_be_completed 

return is a diff of instrument tasks that need to be completed still. tasks granular as possible the diff should be grouped by lifecycle the diff should be ordered by priority of tasks

Module inside of the tad repo. For now the the cli will live inside of the Tad repo Runnable as CLI We start with just an IAMA and "Technische documentatie voor hoog-risico AI-systemen"

$ check-state urn:nl:aivt:ir:td:1.0,urn:nl:aivt:ir:1.0 my-system-card.yaml
probleemanalyse
ontwikkelen, urn:nl:aivt:ir:1.0:2.A.2.2
ontwikkelen, urn:nl:aivt:ir:1.0:2.A.2.3
beheer, urn:nl:aivt:ir:1.0:2.A.2.4

Indication for order of urns:

• lifecycle (group)
• dependencies come earlier
• timestamps of related tasks
• role (implied by urns?)

We need to expand system cards with

• instrument urn
• timestamp

We take the instrument register as is.

Adjustments while building

• when a task has more than one lifecycle, we now pick the first one for which lifecycle/group it belongs to

[uittenbroekrobbert]

Questions we currently have and need to be refined/decided @robbertbos @anneschuth: Regarding: Lib inside of the tad repo and Runnable CLI

Solution A: add a check-state 'command' in scripts, add the function to the main 'AMT' and package everything together. This means 'all of AMT' is needed to use the CLI.

Solution B: if a "pip install check-state" solution is preferred, we may need to split the AMT project into separate 'modules' or projects, like 'shared, cli, web'.

Other solutions may be possible.