Scout Blog
Starlette 1.10.0 cuts form-parsing CPU and memory, stops handled exceptions from pinning request memory, and ships a handful of behavior changes FastAPI teams should read before upgrading.
Symfony 8.1.10 fixes CVE-2026-45070 in the Mime component, which accepted parameter names containing non-token characters, plus a cluster of related mail-hardening fixes. Here's why PHP and Laravel teams should upgrade.
Nokogiri 1.19.5 patches three low-severity security advisories, including a mutation XSS that matters to anyone sanitizing HTML, plus a long run of GC-compaction crash fixes. Here's what Ruby and Rails teams should know.
MongoDB's Node.js driver 7.8.0 fixes a timeoutMS connection-pool bug that could leak other users' data (CVE-2026-119179) and an SRV hostname-verification gap. Here's what changed and why Node teams should upgrade now.
Symfony 5.4.54 and its 6.4, 7.4, and 8.1 siblings fix three high-severity CVEs, all reachable by one unauthenticated request. What each one does, and the exact versions that close them.
Resque 3.2 rejects cross-site requests to resque-web and makes single-job Retry and Remove POST-only. What was exposed, what breaks, and what to check.
Scout is in Claude's connector directory, Scout MCP returns leaner and safer responses, Scout Trails opens its beta, and anomaly charts now show severity.
GoodJob 4.22 deprecates concurrency keys and max_threads ahead of v5. How to migrate to labels, a deploy gotcha, and what throttling looks like in APM.
Connect Scout to Claude in one click and ask about slow endpoints, N+1 queries, errors, and anomalies in plain language, with read-only access.
FastAPI 0.146 to 0.148 change root_path, omitted boolean form fields, and default validation. Here's what each one breaks and what to check first.
Scout Trails gives engineering teams one shared view of AI coding agent activity, including every session, tool call, and dollar spent. It works with Claude Code and other OpenTelemetry-compatible agents, and the beta is free.
What New Relic actually costs a five-engineer team, seven alternatives compared against it, and a step-by-step plan for swapping agents, dashboards, and alerts.