v1.4.8 Available Now

Data Contracts as
Code-Reviewed Guards

Catch schema drift, quality violations, and SLA breaches before they reach production. The open-source data guardian for modern pipelines.

Explore Scenarios
Integrates with:
Airflow Dagster Prefect Snowflake PostgreSQL
bash — dcvpg validate

Engineered for Data Engineers

See how DCVPG solves real-world pipeline nightmares automatically with actionable visualizations.

🛡️

Quarantine Bad Batches

Stop silent failures. When upstream sends nulls or schema changes, DCVPG isolates the bad batch in PostgreSQL and prevents downstream BI dashboards from displaying corrupted metrics.

  • Detect: Identifies schema or value anomalies instantly.
  • Isolate: Diverts faulty rows away from production.
  • Replay: Easily push data back through once the contract is fixed.
Source Data
DCVPG Guard
Production WH
Quarantine Vault
🤖

AI Auto-Healing

When a CRITICAL violation occurs, our LLM agent analyzes the schema drift, proposes a fix to the YAML contract, and opens a GitHub PR automatically for your review.

  • Analyze: Profiles the live table to detect new types.
  • Draft: Adjusts YAML (e.g. Float to String).
  • PR: GitHub integration creates tracked fix requests.
pull_request.md
Update orders_raw YAML: Schema Drift
Detected type change in column 'status' from string -> varchar(50)
- type: string
+ type: varchar(50)
Review & Merge
⏱️

Enforce SLAs

Track row-count anomalies and freshness SLA breaches. If an overnight load produces 10x fewer rows than statistically normal, get alerted instantly via Slack or PagerDuty.

  • Freshness: Guarantee data is no older than N hours.
  • Volume Control: Alert when row volumes fall out of historical bounds.
Mon
Tue
Wed
Thu
SLA BREACH: Row volume -80%
Expected Volume Baseline
💬

MCP Chat Ops

Integrates directly into Claude Desktop or Cursor. Ask questions like "What pipelines are failing?" or "Replay batch abc-123" and manage your platform using natural language.

Replay quarantine batch req-891f2a now that PR 42 is merged.
🤖

Zero-Friction Integration

DCVPG wraps around your existing pipelines. No rewrites needed. You simply add it as a quality gate.

  • 1. Profile: Generate YAML contracts automatically from live tables.
  • 2. Validate: Run checks explicitly in your Airflow DAG or Prefect flow.
  • 3. Guard: Violations are quarantined. Valid data flows to production.
airflow_dag.py
# Add a single operator to your existing DAG
from dcvpg.orchestrators.airflow import DataContractValidatorOperator

validate = DataContractValidatorOperator(
    task_id="validate_orders",
    contract_name="orders_raw",
    config_path="/opt/airflow/dcvpg.config.yaml",
)

# Gate your existing load task
extract >> transform >> validate >> load_to_prod

Documentation Catalog

Explore the complete Data Contract Validator & Pipeline Guardian ecosystem.

contract:
  name: "orders_raw"
  schema:
    - field: "id"
      unique: true

Contract Authoring

Full YAML field reference and capabilities for defining nullability, types, and SLAs.
class NoWeekendOrders(BaseRule):
  def validate(self, data, params):
    # custom logic execution
    return ValidationResult(...)

Custom Rules Extensions

Write generic Python validation extensions to augment the built-in rule sets.
PG
SN
S3
BQ

Database Connectors

Configure internal connections for PostgreSQL, Snowflake, S3, GCS, and more.
What pipelines failed?
1 contract violation found in auth_db.

MCP Server Setup

Integrate directly with Claude Desktop or Cursor for natural language pipeline management.
GET /api/v1/contracts
POST /api/v1/generate
PUT /api/v1/quarantine

REST API endpoints

Expose and consume backend REST services for orchestrators, scripts, and webhook alerts.
🚀
5 Minutes to Prod

Quickstart Guide

Fast track installation, setup, and deploying your very first Data Contract Validator.