Skip to content

Support Engineer

The Support Engineer handles the technical side. It receives error alerts from your monitoring tools, investigates the cause by reading your code, and hands you a diagnosis with a path to the fix.

Supported sources

Section titled “Supported sources”
  • Sentry — connected via OAuth.
  • GitHub Actions — connected via OAuth.
  • Vercel — connected via webhook.

How analysis works

Section titled “How analysis works”

On every alert, the agent reads the source code of the connected repository (via GitHub OAuth), cross-references it with the stack trace and error context, and builds a diagnosis. Instead of just echoing the error message, it points to the likely point of failure.

Ready-to-paste fix prompt

Section titled “Ready-to-paste fix prompt”

Its main output is a fix prompt — text ready for you to paste into your AI coding assistant (like Claude Code) and apply the fix. The agent doesn’t change your code; it prepares the path.

Error context + relevant code snippet + suggested fix
→ paste into your AI assistant → review → apply

Auto-comments on Sentry

Section titled “Auto-comments on Sentry”

Because the audience is internal (your dev team), the agent auto-comments on the Sentry issue with its diagnosis, with no approval required. This is the only Support Engineer flow that skips your sign-off.

Learning loop

Section titled “Learning loop”

When a dev resolves the issue in Sentry, the agent records the outcome and learns from it. Its next analyses of similar errors get sharper.