Managed Mode (Essentials Plan) – LinearB Support

Use this guide to understand how Managed Mode works on the Essentials plan, which AI automations are available, and how to enable or disable them from the LinearB platform—without setting up gitStream config files or runners yourself.

Overview

On the Essentials plan, LinearB AI runs in Managed Mode by default. That means:

LinearB hosts and operates the AI automations for you.
You do not install or maintain gitStream yourself.
You do not need GitHub Actions, CI pipelines, or .cm YAML files.

You control which AI automations are active (for example, AI review, AI descriptions, smart labeling, Dependabot approvals), and LinearB takes care of running them behind the scenes.

Before you begin

Managed Mode currently supports GitHub-only. Other Git providers cannot run AI automations under Managed Mode.

Prerequisites

A LinearB Essentials subscription.
A connected Git provider (for example, GitHub) already integrated with LinearB.
Admin (or equivalent) permissions in LinearB to manage AI services.

GitHub-specific note

If you are using the LinearB GitHub App, you do not need a separate gitStream GitHub App.
If a standalone gitStream GitHub App was previously installed, it should be removed so the LinearB GitHub App is the single integration point.

How Managed Mode works on Essentials

In Managed Mode, LinearB:

Runs AI automations on a LinearB-managed runner, not on your own infrastructure.
Uses the Git metadata from your existing Git integration (no extra setup in your repos).
Applies a curated set of AI automations (see below) based on what you enable in the UI.
Centralizes all AI configuration inside LinearB instead of in repository configuration files.
Managed Mode does not support custom gitStream rules or .cm configurations; only the curated automations shown in the LinearB UI are available.

As an Essentials customer, you always use Managed Mode—but you still have control over which managed automations are turned on and where they apply.

Available Managed Automations

You can turn these automations on or off from the LinearB UI.

AI Code Review

Automatically review introduced changes in a PR with LinearB AI.

Runs on non-draft, non-bot PRs.
Surfaces issues and suggestions directly in the PR.
Helps reviewers focus on what matters most.

AI PR Description

Generate or improve PR descriptions with AI so reviewers get a clear, consistent summary of the change.

Summarizes what changed and why.
Standardizes description quality across teams.
Improves review speed and understanding.

Smart Labeling (Agent Coding Detection)

Automatically label PRs that used AI coding tools (such as Claude Code, GitHub Copilot, Cursor, or LinearB AI).

Detects AI assistance from known patterns and metadata.
Applies labels like Claude Code, GitHub Copilot, Cursor AI, or LinearB AI.
Enables tracking of AI-assisted work and its impact.

Estimated Time to Review (ETR)

Estimate and label how long a PR will take to review.

Calculates an estimated review time (in minutes) based on the PR.
Applies labels like 5 min review, 12 min review, etc.
Uses color or naming to indicate complexity (short vs longer reviews).

Dependabot Minor Bump Auto-Approve

Automatically approve (and optionally auto-merge) Dependabot PRs that perform minor version bumps.

Reduces manual effort on safe, routine dependency updates.
Helps keep dependencies more current with less friction.

Dependabot Patch Bump Auto-Approve

Automatically approve (and optionally auto-merge) Dependabot PRs that perform patch version bumps.

Ensures security and bugfix updates are applied quickly.
Keeps noise low while maintaining safety.

Step-by-step setup

Step 1 – Open AI Services / Managed Mode in LinearB

Log in to the LinearB app as an Admin (or equivalent).
Navigate to the area where you manage AI services or AI settings in LinearB.
Confirm your Git provider (for example, GitHub) is connected.

Step 2 – Choose which Managed automations to enable

In AI Tools, locate the list of available automations:
- AI Code Review
- AI PR Description
- Smart Labeling (Agent Coding Detection)
- Estimated Time to Review
- Dependabot Minor Bump Auto-Approve
- Dependabot Patch Bump Auto-Approve
Toggle on the automations you want to use.
- Many teams start with AI PR Description + Estimated Time to Review to validate noise and value.
- Then add AI Code Review, Smart Labeling, and Dependabot automations as they get comfortable.
Save your changes.

Step 3 – Confirm scope (repos / teams)

Depending on how AI Services are scoped in your account, Managed Mode can be applied globally or more selectively.

Apply automations broadly across all repos connected to LinearB, or
Start with a subset of repos or teams as a pilot, then expand.

Verify

Check that Managed Mode automations are working

Pick a repository connected to LinearB and in scope for your Managed Mode configuration.
Create a new, non-draft PR with some code changes from a human user (not a bot).
Within a short period, confirm:
- An AI review comment appears (if AI Code Review is enabled).
- An AI-generated or AI-updated description appears (if AI PR Description is enabled).
- AI tool labels appear (if Smart Labeling is enabled).
- An estimated review time label appears (if Estimated Time to Review is enabled).
For Dependabot:
- Check a Dependabot PR that performs a minor or patch version bump.
- Verify it is approved (and merged, if configured) when the conditions are met and the related automation is enabled.

Troubleshooting

Problem	Possible cause	How to fix
No AI actions on PRs	Managed Mode automations are not enabled, or the repo/team isn’t in scope	Re-open the AI Services / Managed Mode screen, confirm the desired automations are toggled on, and that the target repos/teams are included.
Draft PRs or bot PRs are ignored	AI automations intentionally skip drafts and bot authors	Test with a non-draft PR authored by a human user.
Dependabot PRs are not auto-approved	Dependabot automations are disabled, or the PR is not a minor/patch bump	Enable the relevant Dependabot automation and verify that the PR’s bump type and naming match what the automation expects.
No AI labels for tools (Claude, Copilot, etc.)	Smart Labeling is disabled or detection patterns not present	Enable Smart Labeling and test with PRs that clearly include the relevant co-author or AI usage patterns.

FAQs

As an Essentials customer, do I need to install gitStream myself?
No. On Essentials, LinearB runs AI automations in Managed Mode for you. You do not need to install or manage gitStream infrastructure.

Can I choose which AI automations to use?
Yes. Managed Mode is always on as the execution model, but you can decide which managed automations (AI Review, AI PR Description, Smart Labeling, ETR, Dependabot automations) are enabled.

Can I switch to self-managed gitStream on Essentials?
No. Essentials uses Managed Mode as the delivery model for AI automations. Self-managed gitStream is available on other plans.

Does Managed Mode modify my code?
No. Managed Mode reads code and metadata to provide comments, labels, and approvals but does not change source files.

Next steps