tables

Detect anomalies in a table

Scan a table for unusual patterns: volume drops/spikes, data gaps, value concentration, high null rates, stale data. Severity-ranked. Tables > 100k rows use a sampled path (~5%) — alerts may carry sampled: true. Quota: 1 capability unit.

get/v1/tables/{t}/anomalies

Request

  • Base URL: https://data.thinair.co
  • URL: https://data.thinair.co/v1/tables/{t}/anomalies
  • Auth: API key in header x-api-key

Path parameters

tstring required

Table name. Accepts table, schema.table, or database.schema.table forms — same wire shape the MCP detect_anomalies tool accepts. Use ?connection= to override the tenant's default connection.

Example:public.telemetry_events

Query parameters

connectionstring

Connection name (display) or id (uuid). Resolved via name match first, then id match. Call /v1/connections to list available names.

Example:fleet
columnstring

Column to focus the scan on. Today maps to the date-trend column for volume / staleness anomalies; the post-migration spec broadens it to a per-column allowlist.

Example:created_at
since_isostring date-time

ISO-8601 cutoff. Anomalies older than this are excluded. Today the underlying scan uses a fixed 14-day window; the value is reserved for the migrated implementation.

Example:2026-04-01T00:00:00Z
severity'low' | 'medium' | 'high'

Minimum severity to include. low returns everything; medium drops low; high drops low and medium.

Example:medium

Response

Anomaly scan result.

summarystring

Changes

No recorded changes to this operation across all 1 revision of this API.