BoltPipeline logo
Platform

Security & Trust at BoltPipeline

Your data never leaves your environment. Security, governance, and auditability are core design principles — not add-ons.

TL;DR for Executives

Your data never leaves your environment. BoltPipeline operates using metadata and execution signals only. Pipelines run inside your database, with customer-owned SQL artifacts, full lineage, approval gates, and audit trails. No proprietary runtime. No vendor lock-in.

Six Security Pillars

Built on a zero-trust, boundary-first security model. Every control is part of the architecture.

Boundary-First Design

  • No raw data ingestion into the control plane
  • No data replication, caching, or external storage
  • Only metadata and execution signals leave your environment
  • All artifacts are ANSI SQL and customer-owned

Identity & Authentication

  • Memory-hard password hashing (Argon2id)
  • Multi-Factor Authentication (TOTP)
  • Short-lived access tokens (asymmetric)
  • Refresh token rotation with replay detection
  • Layered brute-force protection

Per-Agent mTLS

  • Every agent gets a unique X.509 certificate
  • 3-tier PKI hierarchy (Root CA → Intermediate → Agent)
  • Full chain validation on every request
  • Cryptographically bound to tenant and workload

Access Control & Tollgates

  • Role-based access control — Viewer, Developer, Operator, Admin
  • Hard approval tollgates between Dev → Integration → Production
  • Separation of duties — authors cannot approve their own pipelines
  • Re-certification required at every promotion stage
  • Full audit trail of approvals with user identity and timestamps
  • Pluggable SSO support (roadmap)

Platform Security

  • TLS encryption in transit, encryption at rest
  • Hardened service-to-service communication
  • Least privilege across all services
  • Secure defaults with explicit opt-in for elevated capabilities

Compliance & Assurance

  • SOC 2 aligned (access control, monitoring, incident response)
  • ISO/IEC 27001 aligned risk management
  • Clear audit trails and change history
  • Incident response and revocation capabilities

Architecture & Isolation

Customer data stays inside your database and network boundary at all times.

In-Environment Agent

A lightweight Python agent runs inside your infrastructure — your VPC, warehouse, or controlled environment. It executes validations, profiling, and pipeline logic close to the data. Credentials are managed entirely in your environment.

Metadata-Only Control Plane

The Command Center coordinates planning, governance, and visibility using metadata and execution signals only. Table names, column names, schema structure, validation results — never raw row data. Database credentials never collected or transmitted.

BoltPipeline architecture showing agent in customer environment and metadata-only control plane

What Flows. What Stays.

A clear boundary between what leaves your environment and what never does.

Flows to Command Center

  • Table and column names
  • Schema metadata (types, constraints)
  • SQL text (for compilation)
  • Aggregate profiling metrics (counts, %, min/max)
  • Validation results
  • Execution status signals

Never Leaves Your Environment

  • Raw row data
  • Database credentials & passwords
  • PII content (only match counts returned)
  • Query results
  • Business data from your tables
  • Connection strings or secrets

Transparency & Contact

Trust is built through transparency. If you have questions, need documentation, or want to run a security review, reach out directly.

Detailed security architecture, control mappings, and audit evidence are available under NDA as part of procurement reviews.

security@boltpipeline.ai

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.