What's new

Adding a custom MCP server now detects standards-based OAuth automatically and keeps manual credentials as an explicit fallback.

• Server names keep their original capitalization while Superlog generates a protocol-safe identifier behind the scenes.
• Authentication feedback is compact and manual options stay out of the way until you ask for them.

A rebuilt incident page centered on a single, unified activity feed — the investigation, comments, and status changes all read as one timeline.

• When an investigation pauses to ask you something, the question shows up inline instead of getting buried.
• Query cards deep-link straight into Explore with "Open in Explore".
• Log and trace detail now open as a full-screen overlay with room to read.

@-mention the Superlog bot in any channel to ask about your code and telemetry and get an answer in the thread. The bot also self-joins your notification channel, so replies you post on an incident thread reach the investigation agent.

Connect Vercel, Railway, or Render in a couple of clicks and Superlog ingests that platform's logs and metrics for you — no manual OTLP wiring. Joins the existing AWS and Cloudflare connectors.

• Telemetry pages show loading skeletons instead of blank flashes while data streams in.
• The metric-name picker reads from an hourly rollup, so it stays fast on high-volume projects.

A new Install MCP tab in project settings walks you through connecting an MCP client to Superlog, and the settings navigation is flattened so everything is one click away.

Issues now understand real silence and observation windows, and a recurrence chains onto the original incident instead of alerting you from scratch every time it comes back.

The Explore trace list is served from derived rollup tables, so it loads quickly even on projects pushing a high volume of spans.

Added a public privacy policy page, linked from the site footer.

Every contiguous firing of an alert is now grouped into a single episode, so a flapping threshold reads as one event instead of a wall of duplicates.

• Alerts now open to a list of their past episodes.
• Incidents link back to the episode that triggered them.
• The alert editor shows a live preview chart with your threshold drawn on it.

Grafana-style $variable template variables let one dashboard serve many services or environments. Pick a value from the top of the dashboard and every widget re-scopes to it — no more cloning a dashboard per service.

Rebuilt how we store and query exceptions so error views and trace detail load quickly even on high-volume projects. Loading spinners that used to hang on big time ranges now resolve fast.

The investigation agent now remembers what it learns about your codebase and infrastructure across runs. Root-cause patterns, architecture facts, and corrections carry forward, so each investigation starts smarter than the last.

Choose a default org and project that opens whenever you start a new session, instead of landing on whatever you looked at last.

Connected MCP clients can now read and edit a project's issue filter, project context, and agent memories — the same knobs that shape how Superlog investigates issues — without leaving your editor.