Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.anomalyarmor.ai/llms.txt

Use this file to discover all available pages before exploring further.

The /armor:quality skill helps you create and manage data quality checks including metrics and validity rules.

​
Usage

/armor:quality
Or ask naturally:
  • “Add a null check for the email column”
  • “Create a row count metric for orders”
  • “What quality checks exist for this table?”

​
Quality Check Types

​
Metrics

Track quantitative measurements over time:
  • row_count: Number of rows
  • null_rate: Percentage of null values
  • distinct_count: Unique value count
  • freshness: Time since last update

​
Validity Rules

Validate data integrity:
  • NOT_NULL: Column must not contain nulls
  • UNIQUE: Values must be unique
  • ACCEPTED_VALUES: Values must be in allowed list
  • REGEX: Values must match pattern

​
Example Usage

​
Add Null Check

You: Add a null check for the email column in customers

Claude: Created NOT_NULL validity rule:

Rule: email_not_null
Table: warehouse.public.customers
Column: email
Severity: warning

The rule will alert if any null values are found in the email column.

​
Create Row Count Metric

You: Create a daily row count metric for orders

Claude: Created row count metric:

Metric: orders_row_count
Table: warehouse.public.orders
Type: row_count
Interval: daily

Current value: 1,234,567 rows
You'll be able to track trends over time.

​
Check Quality Status

You: What quality checks exist for the orders table?

Claude: Found 5 quality checks for orders:

Metrics:
- row_count: 1,234,567 (passing)
- null_rate on customer_id: 0.1% (passing)
- distinct_count on status: 5 values (passing)

Validity Rules:
- order_id unique (passing)
- customer_id not null (failing - 12 nulls found)

1 failing check needs attention.

​
Common Questions

​
When should I use a metric versus a validity rule?

Use a metric when you want to track a number over time and alert on anomalies (null rate climbing from 0.1% to 5%). Use a validity rule when you want a hard pass/fail check on every run (email must match a regex, status must be in an allowed list).

Profile

Table statistics

Coverage

Find gaps in monitoring