rdt-model-govern

Pulls stewardship metadata (owner, data steward, classification, PII flags, SLA, and terms of use) from Collibra for a given entity. This metadata feeds into rdt-model-compile to produce governance-enriched data contracts.

Access gate — A07

Live Collibra calls are blocked until access task A07 resolves. All calls currently use StubCollibraClient which returns fixture data. The stub is the correct behaviour — do not attempt to configure a live Collibra URL until A07 is marked resolved in GitHub Issues.

Property	Value
Binary	rdt-model-govern
Pipeline phase	2 — Enrich
Network	Collibra REST API (stub until A07)
Credentials	TBD — waiting on A07
Threading	Async
Output	Printed to stdout

Subcommands:

Subcommand	Description	Implemented
pull	Pull governance metadata for an entity	Yes (stub)

Usage - Pull governance metadata (dry-run):

export RDT_TARGET=dev
rdt-model-govern --dry-run pull --entity waste-tracking

Output:

[dry-run] govern pull: entity=waste-tracking
[dry-run]   owner:               Environmental Data Office
[dry-run]   steward:             Maria Schmidt
[dry-run]   data_classification: Internal
[dry-run]   pii:                 false
[dry-run]   sla_days:            1
[dry-run]   terms_of_use:        Restricted to Roche internal use. Not for external distribution.

Usage - Pull governance metadata (standard mode):

export RDT_TARGET=dev
rdt-model-govern pull --entity waste-tracking

Output:

entity:               waste-tracking
owner:                Environmental Data Office
steward:              Maria Schmidt
data_classification:  Internal
pii:                  false
sla_days:             1
terms_of_use:         Restricted to Roche internal use. Not for external distribution.

Global flags:

Flag	Short	Env	Description
--target	-t	RDT_TARGET	Required. dev, test, or prod
--entity	-e	—	Entity slug (required for pull)
--dry-run	-n	—	Print what would be pulled without making API calls
--quiet	-q	—	Suppress [target: dev] header line
--config	-c	—	Path to roche-data.toml (default: ./roche-data.toml)

Configuration: Collibra base URL is configured in roche-data.toml. Leave blank until A07 resolves.

[collibra]
base_url = ""    # TBD — waiting for A07 (issue #25)

Governance metadata fields:

Field	Type	Description
entity_id	string	Entity slug this record belongs to
owner	string	Business owner (team or person responsible for the data product)
steward	string	Data steward (accountable for data quality and governance)
data_classification	string	Roche classification: Public, Internal, Confidential, Restricted
pii	boolean	Whether the entity contains personally identifiable information
sla_days	integer	Freshness SLA in days — data older than this triggers a G3 alert
terms_of_use	string	Usage terms for downstream consumers. Defaults to "Contact data owner for terms of use" when Collibra omits the field

Fixture data: While A07 is unresolved, responses come from: cli/common/src/clients/fixtures/collibra/waste-tracking.json

{
  "entity_id": "waste-tracking",
  "owner": "Environmental Data Office",
  "steward": "Maria Schmidt",
  "data_classification": "Internal",
  "pii": false,
  "sla_days": 1,
  "terms_of_use": "Restricted to Roche internal use. Not for external distribution."
}

To add a fixture for a new entity, create cli/common/src/clients/fixtures/collibra/{entity-slug}.json following the same structure, then add a match arm to StubCollibraClient::load_fixture in cli/common/src/clients/collibra.rs.

Error handling:

no Collibra fixture found for entity

Error: no Collibra fixture found for entity 'new-entity'

The stub client has no fixture for this entity. Add one to cli/common/src/clients/fixtures/collibra/new-entity.json before running.

--entity is required for this command

Error: --entity is required for this command

Pass -e <entity-slug> or --entity <entity-slug>.

[target: dev] printed unexpectedly

This is informational output to stderr. Suppress it with --quiet or -q.

Integration with rdt-model-compile: The governance metadata pulled here feeds directly into rdt-model-compile contract. The compile binary calls StubCollibraClient internally — you do not need to run rdt-model-govern separately as a pipeline step in the current implementation. This module exists as a standalone audit and inspection tool.

When A07 resolves, the compile binary will accept a --governance-path pointing to a governance JSON file exported by this tool.