BoltPipeline logo
Resources

Release Notes

What we've shipped — version by version.

v1.0.0March 2026GA

The first general availability release of BoltPipeline. SQL in, governed pipelines out — with full lifecycle management, security by architecture, and zero access to your raw data.

P

Pipeline Compilation

  • 8-stage compilation: splitting, classification, validation, dependency resolution, SCD generation, lineage, profiling, artifact assembly
  • SQL-to-pipeline — submit SQL with comment-based metadata tags, get deployment-ready artifacts
  • Dependency resolution with topological sort (two-level DAG: group-level + microstep-level)
  • Template-driven code generation via Jinja (dialect-aware, auditable, consistent)
  • Snowflake adapter (GA)
S

SCD Automation

  • SCD Type 0 (append-only) — new rows appended, existing rows never modified
  • SCD Type 1 (overwrite) — MERGE with UPDATE for changed rows
  • SCD Type 2 (full history) — hash-based change detection, effective dates, active/inactive flags, versioning
  • Auto-generated staging tables, merge SQL, and audit columns
  • DB-free generation — SCD logic computed without database connection
L

Column Lineage

  • AST-derived lineage — computed directly from SQL using sqlglot, no runtime tracing
  • Table-level and column-level lineage across all pipeline steps
  • DML coverage: INSERT...SELECT, UPDATE, DELETE, CTAS, CREATE VIEW, MERGE
  • JSON export for external catalog or governance tools
D

Smart Profiling

  • SQL push-down profiling — all queries execute inside your database, only aggregate metrics returned
  • 12 metric categories: uniqueness, null %, cardinality, data type distribution, and more
  • Join inference with cardinality-based type detection (1:1, N:1, 1:N)
  • PII detection — 5 regex patterns (SSN, credit card, email, phone, IP) via SQL REGEXP push-down
  • Pipeline-aware scope — only referenced tables profiled, no warehouse-wide scans
  • Background scheduling — profiling runs during low-activity database windows
  • Production-first design — production profiles serve as source of truth for all environments
V

Validation & Rule Engine

  • 16-rule validation engine — column existence, data types, contracts, audit columns, nullability, SCD configuration, schema drift
  • Auto-generated actionable DDL (ALTER TABLE / CREATE TABLE) to resolve schema drift
  • Custom SQL tests per pipeline step
R

Drift Detection

  • Schema drift — column additions, removals, type changes detected automatically
  • Volume anomaly — row count changes flagged (warning at 50%, critical at 90%)
  • Freshness monitoring — staleness tracking with configurable thresholds
  • PII drift — new PII detected in previously clean tables blocks deployment
  • SCD readiness — key uniqueness and temporal gap validation
A

Open Format Artifacts

  • DML Plan SQL (.sql) — deployment-ready SQL with INSERT, MERGE, SCD logic
  • Actionable DDL (.sql) — ALTER/CREATE statements to resolve schema drift
  • Plan Summary (.yaml) — step list, execution order, SCD types, metrics
  • Airflow YAML (.yaml) — orchestration definition with resolved dependencies
  • Lineage Graph (.json) — table-level and column-level lineage
  • Profiling Results (.json) — column stats, join candidates, drift findings
  • No vendor lock-in — if you leave, you keep every artifact
K

Security

  • Per-agent mTLS — every agent gets its own X.509 certificate with full PKI chain validation
  • Metadata-only communication — raw data and credentials never leave your environment
  • Credential isolation — database passwords managed entirely in your secret manager
  • Scratch schema execution — profiling and validation never modify production tables
  • One-way agent communication — agents pull instructions, platform cannot reach into your infrastructure
  • Auto-rotating certificates — 90-day cycle with session-based refresh
O

PCO Lifecycle & Operations

  • Plan → Certify → Operate governed lifecycle with tollgates between phases
  • DAG-scheduled parallel execution with dependency-aware ordering
  • Offline resilience — encrypted local result queuing with automatic reconnection
  • Environment promotion — Dev → Integration → Production with re-certification at each stage
  • Role-based access — Viewer, Developer, Operator, Admin with separation of duties
  • 6 background task handlers for async pipeline processing
C

Platform

  • Console — web UI for SQL upload, pipeline management, lineage, profiling results
  • Command Center — hosted compilation, validation, artifact storage, agent work queue
  • Agent — lightweight Python runtime in your infrastructure, polls CC for work
  • Multi-tenant architecture with tenant isolation
  • Per-pipeline pricing — all features included, unlimited users

Test Coverage

186

Agent tests

388

Command Center tests

64

Stored procedures

What's Next?

See the full roadmap — upcoming features, multi-DB support, CI/CD integration, and more.

Turn SQL into Production-Ready Data Pipelines — Faster and Safer

SQL-first pipelines, validated and governed — executed directly inside your database.

No new DSLs. No fragile orchestration. Just SQL with built-in validation, lineage, and governance.