Database Change Management (DCM) Projects, now in public preview, bring declarative infrastructure management to Snowflake. And the dcm skill in Cortex Code turns that power into an AI-guided workflow — from first file to first deployment. You describe what your infrastructure should look like, Cortex Code helps you write the definitions, the DCM skill helps you validate them before a single object changes, and DCM handles deployment across your environments.
Schema drift is quiet. Pipelines built on tables managed by ALTER TABLE scripts grow out of sync the moment two engineers touch the same object from different branches. Multi-environment deployments — dev, staging, prod — mean duplicating logic and hoping the config variables stay consistent. And when something breaks, the downstream impact is bigger than it looks.
The traditional fix is rigid: migration scripts, heavy CI/CD tooling, external state stores. These work, but they add friction for every change. Data engineers spend more time tracking what was deployed than building what matters.
DCM Projects solve this with a different model. Instead of scripting changes, you declare what each object should look like using DEFINE syntax. DCM computes the diff, shows you an execution plan before anything runs and applies only what’s needed.
A DCM Project has two core components: a manifest.yml that defines your deployment targets and Jinja2 templating variables, and one or more definition files — standard SQL using DEFINE <object> syntax with variable substitution for each environment. The structure is clean, versioned in Git and deployable with a single CLI command.
DCM Projects supports a wide range of Snowflake objects: tables, views, Dynamic Tables, tasks, warehouses, schemas, databases, roles and grants. More objects are being added regularly. Platform teams can define enterprise-wide standards. Feature teams can own their slice and deploy with confidence.
The DCM Projects skill in Cortex Code extends this further. It’s a domain-specific skill that loads DCM context, syntax references and workflow guardrails into the AI — so Cortex Code knows how to use DCM Projects’ declarative, templatized syntax, understands how to help the user validate changes with DCM ANALYZE and PLAN, and can help guide the user towards DCM-specific best practices.
The skill auto-activates whenever you mention DCM, manifest.yml, snow dcm, DEFINE TABLE or a handful of related keywords — no manual invocation required, but users can also invoke DCM using /dcm.
Here’s where the workflow gets real. Say your team has a deployed DCM project managing a set of Dynamic Tables and Tasks for a daily ingestion pipeline. A schema change is coming — you need to add a new column to a source table that feeds three downstream objects. In the old model, this is a multistep manual process with real risk to downstream objects.
The same DCM commands work in CI/CD. The Snowflake Labs DCM repository ships four reusable GitHub Actions workflows:
Test connections on push
Run snow dcm plan on every pull request to main
Deploy to production on merge
Stage-then-prod promotion with gated approvals
With these in place, every schema change goes through the same review loop automatically — PLAN output visible in the PR, DEPLOY gated on approval. Your Snowflake infrastructure changes now carry the same review discipline as your application code.
DCM Projects, currently in public preview, are easy to try: Visit documentation or get hands-on with the Getting Started developer guide. Use the /dcm skill reference or use the term DCM in your natural language prompt to interact with your DCM Projects conversationally, scaffold your first project in minutes, and start shipping infrastructure changes you can review before they run.
Forward-looking statements
This content contains forward-looking statements, including about our future product offerings, and are not commitments to deliver any product offerings. Actual results and offerings may differ and are subject to known and unknown risks and uncertainties. See our latest 10-Q for more information.
Database Change Management (DCM) Projects, now in public preview, bring declarative infrastructure management to Snowflake. And the dcm skill in Cortex Code turns that power into an AI-guided workflow — from first file to first deployment. You describe what your infrastructure should look like, Cortex Code helps you write the definitions, the DCM skill helps you validate them before a single object changes, and DCM handles deployment across your environments.
Schema drift is quiet. Pipelines built on tables managed by ALTER TABLE scripts grow out of sync the moment two engineers touch the same object from different branches. Multi-environment deployments — dev, staging, prod — mean duplicating logic and hoping the config variables stay consistent. And when something breaks, the downstream impact is bigger than it looks.
The traditional fix is rigid: migration scripts, heavy CI/CD tooling, external state stores. These work, but they add friction for every change. Data engineers spend more time tracking what was deployed than building what matters.
DCM Projects solve this with a different model. Instead of scripting changes, you declare what each object should look like using DEFINE syntax. DCM computes the diff, shows you an execution plan before anything runs and applies only what’s needed.
A DCM Project has two core components: a manifest.yml that defines your deployment targets and Jinja2 templating variables, and one or more definition files — standard SQL using DEFINE <object> syntax with variable substitution for each environment. The structure is clean, versioned in Git and deployable with a single CLI command.
DCM Projects supports a wide range of Snowflake objects: tables, views, Dynamic Tables, tasks, warehouses, schemas, databases, roles and grants. More objects are being added regularly. Platform teams can define enterprise-wide standards. Feature teams can own their slice and deploy with confidence.
The DCM Projects skill in Cortex Code extends this further. It’s a domain-specific skill that loads DCM context, syntax references and workflow guardrails into the AI — so Cortex Code knows how to use DCM Projects’ declarative, templatized syntax, understands how to help the user validate changes with DCM ANALYZE and PLAN, and can help guide the user towards DCM-specific best practices.
The skill auto-activates whenever you mention DCM, manifest.yml, snow dcm, DEFINE TABLE or a handful of related keywords — no manual invocation required, but users can also invoke DCM using /dcm.
Here’s where the workflow gets real. Say your team has a deployed DCM project managing a set of Dynamic Tables and Tasks for a daily ingestion pipeline. A schema change is coming — you need to add a new column to a source table that feeds three downstream objects. In the old model, this is a multistep manual process with real risk to downstream objects.
The same DCM commands work in CI/CD. The Snowflake Labs DCM repository ships four reusable GitHub Actions workflows:
Test connections on push
Run snow dcm plan on every pull request to main
Deploy to production on merge
Stage-then-prod promotion with gated approvals
With these in place, every schema change goes through the same review loop automatically — PLAN output visible in the PR, DEPLOY gated on approval. Your Snowflake infrastructure changes now carry the same review discipline as your application code.
DCM Projects, currently in public preview, are easy to try: Visit documentation or get hands-on with the Getting Started developer guide. Use the /dcm skill reference or use the term DCM in your natural language prompt to interact with your DCM Projects conversationally, scaffold your first project in minutes, and start shipping infrastructure changes you can review before they run.
Forward-looking statements
This content contains forward-looking statements, including about our future product offerings, and are not commitments to deliver any product offerings. Actual results and offerings may differ and are subject to known and unknown risks and uncertainties. See our latest 10-Q for more information.
Snowflake now supports dbt Fusion as a selectable version on dbt Projects, designed to improve compilation times for many builds, […]
Brand relevance is now defined by what we see and what we hear. From the high-energy “vibes” of a short-form […]
Finance needs faster answers, not just better reports For most of my career in FP&A, the work started after the […]
Our services help both the big and small organisations in the identification of security breaches and provide outclass solutions to prevent any potential harm.