kontra 0.5.0

Developer-first data quality engine

Kontra

Data quality validation for developers.

Kontra validates datasets against declarative contracts. Define rules in YAML, run them against Parquet, CSV, PostgreSQL, or SQL Server. Get violation counts back.

pip install kontra

30-Second Example

# Profile your data
kontra profile data.parquet

# Draft a starting contract
kontra profile data.parquet --draft > contract.yml

# Validate
kontra validate contract.yml

Output:

PASSED - data.parquet (4 rules)
  COL:user_id:not_null      ✓
  COL:email:unique          ✓
  COL:status:allowed_values ✓
  COL:age:range             ✓

What You Write

# contract.yml
name: users_quality
datasource: data/users.parquet

rules:
  - name: not_null
    params: { column: user_id }

  - name: unique
    params: { column: email }

  - name: allowed_values
    params:
      column: status
      values: [active, inactive, pending]

What You Get

• 13 built-in rules: not_null, unique, range, regex, freshness, and more
• Fast execution: Metadata analysis and SQL pushdown before loading data
• Multiple sources: Parquet, CSV, PostgreSQL, SQL Server, S3
• Python API: Use as a library with kontra.validate(df, rules=[...])
• State tracking: Compare runs over time with kontra diff

Documentation

Doc	Audience
Getting Started	New users
Python API	Library users
Rules Reference	Everyone
Configuration	Project setup
Advanced Topics	Agents, state, execution model
Architecture	Contributors

License

MIT