Tags
Tags are lightweight, free-form classification labels that can be attached to an OORS message and to individual results. The structure is shared across the Bitol standards and follows the same shape everywhere it appears.
What tags are for
Tags are intended for categorization — making observations easier to find, filter, group, and route through tooling. Common uses include:
- Domain or topic:
finance,marketing,iot - Sensitivity / compliance:
pii,phi,gdpr,pci,sensitive - Lifecycle:
experimental,deprecated,terminal - Audience:
internal,external,partner - Operational hint:
hot,cold,archive
Tags are deliberately not a controlled vocabulary: your organisation defines what each tag means. For vendor- or organisation-specific metadata that carries a value rather than a label, prefer custom properties.
Structure
A tags value is always an array of strings. Empty arrays are allowed; duplicate strings within the same array should be avoided.
The same structure applies wherever tags is permitted.
Where tags can appear
tags is available in the following places:
| Location | Notes |
|---|---|
| Top-level (the message) | Tags applied to the observation run as a whole. See Fundamentals. |
| Result | Tags applied to an individual check/metric. See Results. |
Recommendations
- Lowercase, snake_case or single words. Mixed casing makes filtering brittle.
- Be consistent. A tag
piiin one message andPIIin another defeats the purpose. Document your tag taxonomy organisation-wide. - Prefer specific over generic.
customer_piiis more useful thansensitiveonce you have many messages. - Avoid encoding hierarchy in tag strings.
finance/regulatoryis harder for tooling than two separate tagsfinance,regulatory. - Keep tags small in number. A handful per element is plenty; long tag lists become noise.