# CohortsConfig

Cohort-identity + space-aggregation configuration.

**Type**: `object`

## Properties

- name: identifyingAttributes (optional)
  type: array of string
  description: Level-prefixed attribute keys that group objects into cohorts.

- name: metadataAttributes (optional)
  type: array of string
  description: Level-prefixed attribute keys describing cohorts.

- name: numShardBits (optional)
  type: integer
  description: Parallelism multiplier for cohort outputs (anomaly rules ignore this). Splits each cohort across 2^n CohortCalc instances by the low n bits of tsIdLo, trading parallelism for partial per-shard aggregation. 0 = one instance per cohort.

- name: outputs (optional)
  type: array of `CohortOutputDecl`
  description: Ordered cohort-output entries (first-match-wins).