A code review sits in Gerrit, waiting. Tests have passed. The build looks clean, but no one wants to push “submit” at midnight. Then Gerrit Lambda steps in and finishes the job, exactly as you described it, without crossing a security line or waking anyone up.
At its core, Gerrit handles the human side of code review: patch sets, approvals, and auditing. AWS Lambda handles the automated side: executing functions in response to events. Combining the two turns approvals into triggers, reviews into automation, and CI gates into intelligent workflows. Gerrit Lambda links code changes directly to action.
In practice, it works like this: Gerrit emits events on review updates or merges. A Lambda function subscribes to those events through an HTTP endpoint or a message broker like SNS. Once a patch hits a certain label threshold or merges to main, Lambda handles the follow-up—tagging Docker images, triggering Terraform plans, or notifying compliance systems. No polling, no manual scripts, just defined policy triggering compute on demand.
Setting up Gerrit Lambda usually means defining event subscriptions with scoped credentials, storing them securely in AWS Secrets Manager or using roles via OIDC to avoid keys altogether. Permissions map cleanly through IAM policies, limiting each function to one job. If errors happen, CloudWatch provides logs right away, making traceability clear without digging through Jenkins spaghetti.
Typical benefits you can expect:
- Merges trigger instantly, speeding feedback loops by minutes or hours.
- Access policies stay centralized and auditable under IAM or OIDC.
- No persistent servers or daemons, cutting infrastructure drift.
- Review-to-deploy latency shrinks, improving developer velocity.
- Stronger guardrails through code-defined triggers instead of human click paths.
For developer experience, this matters more than it sounds. Gerrit Lambda turns approvals into events instead of chores. You stop copy-pasting pipeline steps and start treating reviews as automation gates. Fewer manual merges, fewer Slack pings asking “who deployed this?” and faster onboarding for new engineers who just need to understand labels and events, not half a dozen build scripts.