Changelog

Flow Observability

Track pipeline runs, throughput, and lineage directly inside the Cascade workspace

• New observability dashboard to monitor every pipeline across your environment
• View run history, success rates, failure trends, and average durations
• Filter by status, environment, or time window with live auto-refresh
• Analytics view with execution heatmaps and slowest-stage insights
• Visualize weekly throughput patterns and export historical run data
• Wired directly into the Cascade Projects workspace

Verify Pipelines

The first major upgrade to the Cascade verifiers. Cascade can now simulate pipelines in memory with full connector and warehouse behavior, no execution required.

• Simulate pipeline runs in memory without touching production infrastructure
• Validate whether a pipeline will hold up in production using real logic, not mock stubs
• Supports Postgres, BigQuery, Snowflake, S3, Python, and custom connectors
• Create in-memory replicas of tables to test loads and transforms safely
• Simulate message passing and monkey-patch task execution at runtime
• Feed simulation results back into the agent for sharper planning

Pipeline Cloud IDE

Every pipeline now gets its own full workspace with editing, preview, and version control

• Manage your entire pipeline project in a dedicated workspace
• Edit definitions inline with syntax highlighting and multi-diff support
• Sync changes to Git, discard edits, or open pull requests
• Access metadata, schedules, and dependency graphs in one place
• Works like a cloud IDE purpose-built for pipeline development

Isolated Secrets Vault

Each Cascade instance now runs its own isolated vault for managing sensitive credentials

• Dedicated secrets vault provisioned per deployment
• Secure, isolated storage for database and cloud credentials
• Auto-fetch and inject secrets into agent workflows at runtime
• Improved compliance and end-to-end auditability

Improved Inline Edits

Inline editing is now more accurate for complex models and can apply many more diffs at once

• Apply up to 12 diffs per edit cycle
• Highlight and preview changes before applying them
• Handle nested CTEs, macros, and long templating logic
• Cut turnaround time through side chat or direct edits

Manual Edits

Blend agent generation with hand-tuned control over every stage of a pipeline

• Inline syntax highlighting and validation still apply to your edits
• Save and sync your changes to S3 or your Git repo as usual
• Combine agent generation with hand-tuned adjustments for full flexibility

Live Previews for Models

Cascade now streams live previews of models as table outputs in the UI, driven by the agent

• Agent generates compiled model previews in table format in real time
• Streams sample result rows directly into an interactive UI table

GitHub and GitLab PR Integration

Cascade now lets you create and manage pull requests on GitHub and GitLab for your pipeline changes

• Create pull requests for pipeline updates directly from the Cascade UI
• Preview the diff of pipeline changes before opening a PR
• Select target repository, branch, and reviewers for GitHub and GitLab
• Track PR status and merge progress within Cascade
• Automatically include pipeline metadata and test results in the PR description