Choose what is looked for

complydoc looks for about forty kinds of identifier: bank details, tax numbers from a dozen countries, names, addresses, and more. Not all of them matter to every folder. A UK insurer has no use for US ZIP codes, and a team that finds organisation names noisy may want them called less serious. The categories file changes this without copying complydoc’s settings.

.complydoc-categories.yaml

schema_version: 1
categories:
  us_zip_code:
    enabled: false
  organisation_name:
    severity: low

Only what differs from complydoc’s own settings is written. A category left out of the file is looked for as it ships, so the file doesn’t drift from the defaults as they change.

Field	Meaning
enabled	false stops the category being reported: no finding, no count, no check. Its values are still masked.
severity	low, medium or high: how serious a finding of the category is called. The policy, complydoc check and the report’s counts follow it.

An audit reads .complydoc-categories.yaml at the top of the folder it audits, or the file --categories names. A pipeline observed with cd.observe takes its categories from the config it is given.

What the report says

A category that is switched off shows nothing, which reads the same as a clean one. So the report says it is not reported:

• the limitations list “Categories switched off”, and the Security page names them above the findings, beside any category that couldn’t run for another reason, such as a missing name model
• the categories section of the JSON lists each change with the shipped setting beside it

A name in the file that is no category, from a typo say, changes nothing and doesn’t stop the run. The report names it.

Change it in the viewer

Under complydoc ui, the Settings page lists every category with a box to switch it off, its severity, and a button to put it back as shipped. A category changed from its default is marked. The file is the one the command line reads; changes apply from the next run. A report opened any other way shows what its run changed, read only.

A category switched off is still masked

Switching a category off takes it out of the findings, the counts and the checks. It does not take it out of the masking: the run still looks for it, so that a value of it is covered wherever the report shows text. A report that stopped looking would show those values as they are written, to everyone the report is shared with.

That means switching off a category that a model reads, such as names, does not make a run quicker.

What this doesn’t do

• Shipped categories can’t be deleted or have their patterns edited here. Switch one off, and add your own concept for what you want instead.
• Concepts are in their own file and have their own severity.