data-move

What This Skill Does

The data-move skill provides a comprehensive, risk-aware framework for executing complex data migrations. Rather than treating migration as a simple database dump, it enforces an engineering-grade ETL workflow. It mandates the definition of clear invariants, field-level mapping, and idempotent batching to ensure data integrity during transitions between databases, regions, or tenants. The skill is designed to prevent silent data corruption and ordering errors by requiring validation, checksums, and shadow-mode comparisons before final cutover. It manages the entire lifecycle of a move, from scope definition to post-cutover reconciliation.

Installation

Install the skill via the OpenClaw CLI: clawhub install openclaw/skills/skills/clawkk/data-move

Use Cases

• Moving a production tenant from a legacy SQL database to a distributed NoSQL store without downtime.
• Executing a massive historical data backfill after a breaking schema migration.
• Implementing a blue/green deployment strategy for data infrastructure where you swap storage backends while maintaining traffic flow.
• Reconciling data divergence between two distributed systems using checksums and business-logic validation.

Example Prompts

1. "I need to migrate 5 million user records from our legacy PostgreSQL cluster to a new sharded instance. Help me draft an invariant list and a strategy for idempotent batching that preserves primary keys."
2. "We are planning a zero-downtime cutover for our order history table. Can you build a rollback runbook based on our blue/green table schema and define the validation rules for a shadow-read verification?"
3. "Our recent migration shows a 0.01% discrepancy in account balances. Use the data-move framework to generate a reconciliation job that identifies the missing entries between the source and target."

Tips & Limitations

• Always start with the scope definition; failing to identify foreign key dependencies early is the primary cause of migration failure.
• Utilize the shadow-read feature in Stage 4 to compare results in real-time before shifting production traffic.
• Ensure your batching implementation includes robust logging to handle potential resume-after-failure scenarios.
• This skill is highly opinionated regarding documentation; if you skip the mapping document stage, the automated validation scripts will not be generated.