Foreman

What it does

• Several stories at once

  Give Foreman a pool of sandboxes and it builds one story in each, side by side. With scratch orgs, every story gets a fresh org of its own. Finished stories release one at a time, so they never trip over each other.

• Fast by design

  Each release deploys only what the story changed, not the whole project, and the dry run runs the story's tests and the ones its change touches, not every test in the org. Every sf call has a time limit, so a stuck command never holds up the line.

• Quality checked at every step

  Builders write the tests first: positive, negative, bulk and null cases. A separate tester that never edits re-runs them in the org, checks coverage per class (75% floor, 90% target by default), and runs LWC Jest and Code Analyzer. Then a Salesforce code reviewer, a compliance reviewer and a security reviewer check the change, and anything they find goes back to the builder.

• Agents that know Salesforce

  Admin work goes to a builder that writes Flows as metadata and puts field access in permission sets. Apex and LWC go to a developer, and OmniStudio and Agentforce to builders of their own. The security reviewer checks Salesforce's own risks: object and field access, sharing and SOQL injection.

• Three ways to release

  Foreman deploys with sf itself. Or your own GitHub Actions job deploys and Foreman confirms it in the org. Or Gearset promotes: Foreman merges the promotion pull request once Gearset passes it, or waits for your Promote click.

• OmniStudio

  Foreman tells the standard runtime from the managed package. It deploys Data Mappers, then Integration Procedures, then OmniScripts and FlexCards, checks the right versions are active, and calls each changed Integration Procedure with a test input. On the managed package it parks the story with the activation steps for you.

• Agentforce

  An Agentforce builder writes the agent, its actions and their Apex or Flows, plus a test spec. Foreman previews the agent with simulated actions, publishes and activates it, then runs the tests. Wrong routing blocks the story, and rolling back reactivates the previous version.

• Rollback where it is safe

  On the first stage a failed check rolls back by itself: Foreman reverts the merge and deploys the reverse delta, metadata only, never data. A failure on a later stage like UAT, which people share, halts and alerts you instead.

• Production stays yours

  Foreman deploys only to orgs on your allow-list, and checks each one is a sandbox before every step. Production can be connected as your Dev Hub or for reading, and releasing to it stays with you.