The database operations layer for AI-native engineering teams. Versioned migrations, immutable audit logs and real-time schema intelligence for Apache Cassandra, Google Cloud Spanner, Google Bigtable, Amazon DynamoDB, TiDB, CockroachDB, YugabyteDB, Vitess, and SingleStore. One platform. Nine databases. Zero schema drift.
Each tool is purpose-built for its database — no leaky abstractions, no lowest-common-denominator API. All share the same CLI contract, the same history table schema, and the same audit log format. SQL, NoSQL, NewSQL, wide-column, distributed OLAP — every major distributed database category covered.
Your AI assistant is fast everywhere except the database. Schema mistakes on distributed systems are catastrophic, expensive to reverse, and impossible to audit after the fact. SchemaShifts closes that gap — giving Claude, Copilot, and Cursor native access to migration state, schema health, capacity risk, and compliance history across your entire data infrastructure.
CQL-native schema migration for Cassandra and every Cassandra-compatible database. Versioned scripts, LWT distributed locking, multi-keyspace deploys, and an MCP server for AI-assisted development.
SQL DDL migration tool purpose-built for Cloud Spanner. Handles Spanner's async DDL operations, interleaved tables, TTL policies, VECTOR indexes, and property graphs — with a full history and audit trail.
YAML-driven schema migration for Bigtable — tables, column families, GC policies, and app profiles. Unique column registry for schemaless governance, drift detection across 9 categories.
The only governance tool that treats DynamoDB schema as code. GSI lifecycle management, TTL policies, table class, streams — versioned, drift-detected, and audited. Other tools have no DynamoDB support.
First-class TiDB migration tool — goes beyond SQL DDL to version TiFlash replicas, placement policies, and resource groups natively in YAML. Handles Serverless tier feature gaps with PARTIAL status tracking.
The only migration tool that manages CockroachDB multi-region topology as code. REGIONAL BY ROW locality, survival goals, zone configs — versioned and drift-detected. Includes DDL job monitoring other tools ignore.
The only migration tool that manages YugabyteDB's colocation, geo-partitioned tablespaces, and xCluster DDL sequencing as code. Goes beyond psycopg2 to handle YB-native YAML migrations other tools ignore entirely.
Vitess-native migration tool — versions VSchema (vindexes, sharding keys, sequence tables) as YAML migrations. Polls Online DDL per shard, detects vtcombo for local dev, and keeps migration history in a dedicated keyspace.
The first migration tool built around SingleStore's distributed SQL model — versions SHARD KEY, SORT KEY (columnstore), and VECTOR INDEX as YAML. 7-rule analyser catches sharding anti-patterns other tools can't even see.
When a SOC2 auditor or HIPAA assessor asks who changed the schema, when, and whether it was approved — cassachange, spanchange, and bigtablechange answer automatically. No additional tooling. No manual evidence gathering.
Every deploy, rollback, script execution, and lock event is written append-only to the history database. Rows are never updated or deleted. Operator identity, timestamp, run ID, and outcome — captured automatically on every migration.
Passwords, SSL certificates, and AstraDB secure bundles are pulled from HashiCorp Vault, AWS Secrets Manager, or Azure Key Vault at runtime. Nothing sensitive touches disk on a CI/CD runner. URIs never appear in logs.
Every schema change is a versioned, checksummed file committed to source control before it runs. No ad-hoc changes can be applied without appearing in migration history. Deploy fencing with tags provides release-level traceability.
Every versioned migration ships with a paired undo script. Any change can be reversed in a single command with a full audit record. This is a documented, tested recovery procedure — exactly what incident response controls require.
Encode naming conventions, mandatory TTL on sensitive tables, prohibited patterns, and PII column requirements as version-controlled rules. Enforced as CI gates with GitHub annotation output. Automated policy enforcement is accepted audit evidence.
Distributed locking (Cassandra LWT, Spanner serialisable transactions, Bigtable advisory lock) prevents two operators or CI pipelines from running simultaneously. No partial schema state. No conflicts. Availability controls satisfied.
| Timestamp | Run ID | Event | Operator | Script | Detail |
|---|---|---|---|---|---|
| 2026-03-23 09:14:02 | 9e2c8f1a | DEPLOY_START | ci-runner-07 | — | tag=2.1.0 dry_run=false |
| 2026-03-23 09:14:03 | 9e2c8f1a | SCRIPT_RUN | ci-runner-07 | V1.0.0__create_users.cql | 44ms SUCCESS |
| 2026-03-23 09:14:03 | 9e2c8f1a | SCRIPT_RUN | ci-runner-07 | V1.1.0__add_orders.cql | 29ms SUCCESS |
| 2026-03-23 09:14:04 | 9e2c8f1a | DEPLOY_END | ci-runner-07 | — | errors=0 tag=2.1.0 |
The enterprise edition ships a 23-rule schema analyser, a CI/CD contract enforcement gate, and a real-time CI/CD analytics dashboard — all surfaced natively inside Claude, Copilot, and Cursor via the MCP server. Catch anti-patterns before they reach production, enforce policy as code, and monitor deploy health across every keyspace.
Every table in every keyspace is scored against 23 rules covering partition hotspots, missing clustering columns, absent temporal anchors, TTL gaps, and ALLOW FILTERING risks. Each finding includes actionable CQL advice — not just a score.
Encode your team's schema rules as version-controlled contracts — naming conventions, mandatory TTL, prohibited anti-patterns, PII column requirements. Enforced as a hard CI gate with exit code 2 on failure. Fully automated, fully auditable.
The MCP server surfaces a real-time CI/CD analytics dashboard — total deploys, success rates, p95 deploy times, slowest scripts per keyspace, and a per-keyspace breakdown. Ask Claude for your deploy health and get a live visual answer.
Every Schemashifts tool ships with a Model Context Protocol server — letting Claude, Cursor, and any MCP-compatible AI assistant query your schema history, run drift detection, validate migrations, and plan deployments without leaving the chat. Schema governance meets agentic AI.
Connect your AI assistant to any Schemashifts tool and unlock schema-aware conversations — ask about pending migrations, get drift reports, validate scripts before running them, or generate new migration files from a natural language description.
MCP servers for dynamochange, tidbchange, cockroachchange, ybchange, vitesschange, and singlestorechange are in active development. cassachange, spanchange, and bigtablechange MCP servers are available now. All follow the same tool schema and Claude Desktop configuration pattern.
The core migration engine is free and open source. The enterprise edition adds compliance infrastructure, schema intelligence, observability, and enterprise platform support.
Everything you need to run reliable schema migrations across Cassandra, Spanner, Bigtable, DynamoDB, TiDB, and CockroachDB in production.
Everything in Community, plus compliance infrastructure, schema intelligence, observability, and support for Google Bigtable proxy, Azure CosmosDB, and Azure Keyspaces.