On this page

For AI agents: a documentation index is available at /llms.txt. Append .md to any page URL for markdown, or send Accept: text/markdown.

Smart Gate

Smart Gate reviews pull requests in your GitHub repositories, suggests a feature gate for risky changes, wraps the change in a Statsig gate when you opt in, then ramps, reverts, and cleans up the gate for you.

Early Access

This feature is in Early Access. During this time, aspects of the functionality may still be developed, and this documentation may not always be up to date. If you have any questions, contact Statsig Support.

Smart Gate is an AI agent that reviews pull requests in your connected GitHub repositories and flags changes risky enough to ship behind a feature gate. When the PR author opts in, Smart Gate opens a second pull request that wraps the change in a new Statsig feature gate. After the code deploys, Statsig ramps the gate on a schedule, reverts it to 0% if a safeguard alert fires, and opens a cleanup PR once the gate has been at 100% long enough.

Use Smart Gate when you want risky changes to ship behind a gate without hand-writing gate checks, picking metrics, or babysitting a rollout. Create a gate yourself with Create a feature gate when you need custom targeting, a permanent gate, or an experiment. If you already manage gate rollouts with Release Pipelines, Smart Gate leaves those gates alone and doesn't ramp them.

Availability

Smart Gate is in early access. Statsig enables it per organization. To request access, contact your Statsig account team or Statsig support.

How Smart Gate works

Nothing in your code changes until someone opts in on the pull request.

Smart Gate understands how your codebase already uses Statsig, so the gate it adds reads like code your team wrote: same SDK, same patterns, same style.

Smart Gate keeps the PR author updated in a comment on the PR, and optionally in Slack. For what PR authors see and do, go to Work with Smart Gate on a pull request.

The gates Smart Gate creates

  • Off in production, on in development and staging: you can test the new behavior before any user gets it.
  • Monitored: each gate gets metrics and safeguard alerts from the project settings you choose.
  • Editable: you can change a gate's rules, metrics, and alerts from the console like any other gate.

What Smart Gate doesn't do

  • Change code without consent: Smart Gate only opens a wrap PR after someone on your team opts in.
  • Block your merge: Smart Gate's check on the PR never fails.

How Smart Gate handles your code

Smart Gate works on your code in an isolated, short-lived environment, and Statsig and its LLM sub-processors don't use your data to train models. For details, go to AI governance, security, and privacy.

Get started

To connect GitHub and turn on Smart Gate, go to Set up Smart Gate.

Was this helpful?