Set up the Dagster+ GitHub Actions or GitLab CI integration by adding the dagster-cloud CI/CD workflow file to the repository
Define a dagster_cloud.yaml manifest listing your code locations and their Docker image references; branch deployments are created from this manifest on each PR
The CI workflow builds the Docker image, pushes it to a registry, and uses the Dagster+ agent to spin up an ephemeral branch deployment scoped to the PR
In the branch deployment, validate the asset graph using dagster-cloud ci check which runs schema validation and detects broken asset dependencies before merge
Configure status checks in GitHub to block merges until the Dagster+ CI check passes, ensuring no broken asset graphs reach production
Tear down the branch deployment automatically when the PR is closed or merged; Dagster+ handles cleanup via the GitHub App integration
Known gotchas
Branch deployments share the same underlying Dagster+ agent as production; resource-intensive test materializations in a branch deployment can starve production runs unless separate agent pools are configured
The dagster_cloud.yaml image tag must reference the exact image built in the same CI run — using 'latest' causes branch deployments to run stale code silently
Branch deployment environment variables must be explicitly copied from the production code location settings; they are not inherited automatically
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp