Sentry

Seer's agent reached open beta, its Claude fix integration went GA, and Monitors and Alerts split into separate, independently-controlled features.

Seer's AI debugging loop expanded across editors. Seer Agent — ask anything about your app, answered from connected telemetry — opened to all users in beta.¹ The Claude integration for automated fixes reached GA; a GitHub Copilot handoff that sends Seer's root-cause analysis to Copilot entered open beta.²³ AI Conversations, a readable log of past agent sessions, also opened in beta.⁴

Monitors and Alerts became separate controls. The old single alerting concept now splits tracking (Monitors) from team notification (Alerts), each managed independently.⁵ Both reached GA in May.

Application Metrics and Cross-Event Querying went GA. Metrics carry high-cardinality tags — user ID, region, plan type — with click-through to the connected trace, log, or error.⁶ Cross-Event Querying joins logs, metrics, and spans in one query.⁷

Snapshots entered early access for visual regression testing, handling image storage, diffing, and CI checks.⁸ The Perforce P4 integration reached GA with stack-trace linking, suspect commits, and code review.⁹

SDKs added gen_ai span streaming. Python 2.60.0 and JavaScript 10.53.0 send gen_ai spans as separate envelope items, sidestepping payload-size limits on AI-heavy workloads; the Hono and Nitro SDKs entered beta.¹⁰ CLI 3.4.3 shipped security fixes — Xcode Info.plist preprocessing off by default, shell-escaped bash-hook arguments, and checksum verification on sentry-cli update.¹¹