Documentation
Last updated 6 September 2026
Frontier finds the downstream business entities that may be stale after a dbt SQL change or a source data change, then proves a targeted recompute against a full reference. Warehouse rows and Snowflake passwords never leave your environment. Billing is not currently enabled.
Start
Quick start
Install the runner from PyPI and complete a first SQL-change assessment.
Concepts
How Frontier works
What Frontier does, what it does not replace, and the assessment lifecycle.
Setup
GitHub installation
Generate a pinned PyPI workflow and store GitHub Actions secrets.
Setup
Snowflake permissions
Least-privilege grants for PR assessment, not ACCOUNTADMIN.
Setup
Semantic manifest basics
Discover a draft mapping, review it, then activate it in Frontier.
Workflows
SQL-change assessments
Compare compiled base and PR SQL, then prove a targeted recompute.
Workflows
CDC data-change assessments
Inspect, capture, prove, and upload Snowflake Stream batches separately from PRs.
Reference
CLI reference
Every current frontier-runner 0.1.1 command, flag, and artifact.
Reference
Understanding an assessment
Field-by-field guide to the PR comment and run-detail page.
Reference
Configuration reference
Flags, environment variables, keychain, config files, and CI secrets.
Reference
Troubleshooting
Stable error codes from install, auth, manifests, CI, Snowflake, and CDC.
Security
Architecture and trust boundaries
Where the runner executes, what SaaS stores, and how manifests are pinned.
Security
Security model
Warehouse rows and credentials stay in your environment.