This month is about shortening the path from detecting a problem to fixing it. You can now create GitHub or Linear issues directly from Scout, Node.js now has complete logs and error monitoring, and Anomaly Detection gets more control over what actually triggers an alert. Here’s what’s new in July:
Node.js: Full Logs and Error Coverage
Node.js now has complete logs and error monitoring. Every exception includes the request trace, timing breakdown, and surrounding log lines.
If you onboarded a Node.js app in June, logs are now available with no reinstall required.
One-Click Issue Creation
Found a bug? Create a GitHub or Linear issue directly from Scout.
Trace detail pages (endpoints, jobs, and the v2 timeline view) and the error detail view now include a Create Issue dropdown. It opens a pre-filled GitHub or Linear issue with the title and a link back to Scout already included. If GitHub isn’t configured yet, you’ll see a setup link instead. Linear works with no additional setup.
Anomaly Detection: Trigger Smarter Alerts
Reduce noisy alerts with direction filtering. Monitors can now detect spikes, drops, or both, either individually or as an organization-wide default.
A traffic drop on a background job queue and a latency spike on a checkout endpoint are different failure modes, and treating them the same creates unnecessary noise. Direction filtering lets you tune each monitor to the behavior you actually care about.
The panel has also been redesigned with inset rows and inline direction controls, so adjusting sensitivity and direction happens in the same view where you’re already reviewing anomaly history.
Anomaly Detection is currently available by request. Reach out to us if you want in. We would love your feedback as we continue to make enhancements.
Ruby Agent Improvements
Two Ruby agent issues were addressed this month, both narrow but worth flagging if they match your setup:
- GraphQL layer tracking now handles unhandled exceptions that skip
end_execute_fieldcorrectly, so a raised exception inside a GraphQL resolver no longer leaves an orphaned trace layer. - Heroku git revision detection now reads
HEROKU_BUILD_COMMIT, which is more reliable across recent Heroku build configurations than the previous detection method.
From the Scout Community
Your AI Coding Agent Is Flying Blind in Production AI coding agents read your codebase well but have no visibility into what your code actually does once it ships. This post walks through connecting Scout’s MCP server so an agent like Claude Code or Cursor can pull error groups, endpoint traces, and N+1 insights directly, turning “the agent guesses at a fix” into “the agent traces the problem to the transaction holding the lock.”
Unified Logs, Traces, and Errors: Why One Tool Beats Three Running separate tools for errors, traces, and logs means manually reassembling a timeline by hand every time something breaks, matching timestamps and request IDs across three UIs. This post breaks down why that correlation is hard to retrofit after the fact, and why tools built with unified observability from the start do not have the problem in the first place.