Over the past couple of months, we focused on making Scout easier to get started with and ready for what's next. That includes future-proofing work like Ruby 4 support, as well as practical improvements that make debugging and operating your apps simpler day to day.
Ruby Support
- The Ruby agent now supports Ruby 4. When you're ready to upgrade, Scout will just work.
- We also added configurable `backtrace_additional_directories` so you can include shared gems, engines, or internal libraries in error backtraces. Let's say you have a complex application with code spread across multiple directories, now you can tell Scout to capture the full context when something breaks.
- Sidekiq visibility improved. We now capture job parameters outside ActiveJob contexts. If you rely on Sidekiq directly, you should see clearer, more useful context when tracking down job-related issues.
- Sample rate handling is more intelligent. Invalid values coerce to sensible defaults while preserving intentional settings, making configuration more straightforward.
Released in: Ruby agent 6.0.2
Python Support
- Error search improved for Python. Problem groups now index and match error locations more accurately and reliably.
Released in: Python agent 3.5.2
Platform Reliability
- We've strengthened platform security practices and improved reliability across the platform. This work doesn't change how you use Scout, but it makes the platform more secure, resilient, and dependable under the hood.
- We also refreshed the signup flow and made promo codes easier to apply during registration. Getting started should be straightforward.
From the Scout Community
- How AutoInstruments uses Prism to parse and modify Ruby source:
Ruby 3.4 made Prism the default parser. We published a detailed look at how Scout's AutoInstruments feature leverages Prism and the Abstract Syntax Tree to automatically trace your custom Ruby code. The post walks through YARV instruction sequences, the `load_iseq` hook, and how we inject timing instrumentation without modifying your source files. Read the deep dive - ShakaCode cut infrastructure costs 90% during Kubernetes migration
ShakaCode migrated their product HiChee from Heroku to Kubernetes on Hetzner, reducing costs by 90%. The challenge: when compute and database layers move, latency patterns shift. Scout helped them see where time was actually going—code, queries, or external calls—so they could distinguish between "the app is slower" and "database latency increased." See how they did it
Try the Scout MCP Server
Your AI assistant can now query Scout performance and error data directly.
Install the MCP server and try prompting your agent with:
- “Why did error rates spike yesterday?”
- “Show me the slowest endpoint last week.”
- “Show me the highest-frequency errors for app Foo in the last 24 hours. Get the latest error detail, examine the backtrace and suggest a fix.
Beyond queries: automated workflows
One team is using the MCP server to automatically identify performance regressions and generate PRs. Their workflow:
1. Query Scout for endpoints that degraded after a deploy
2. Pull the specific slow traces and database queries
3. Ask AI assistant to analyze the bottleneck and suggest fixes
4. Generate a PR with the optimization
Fast, clear answers right where you work. It’s AI-native monitoring for AI-native development.
Give our new Error Monitoring and the MCP server a spin, and let us know what you find useful. We're curious.
Happy Scouting,
Sarah
Product @Scout Monitoring
.png)
