Features

One timeline, from the exception to the page

Error monitoring, performance, uptime, on-call and status pages are the same product here, so an incident does not have to be reassembled from three vendors' dashboards while it is happening.

Error monitoring

Events are fingerprinted into issues by a versioned grouping algorithm, so the same fault stays one issue across deploys instead of splitting every time a line number moves. Variable parts of a message — ids, timestamps, paths — are normalised out before the hash.

Read the docs

01.1Versioned grouping: changing the algorithm never splits your existing issues
01.2Source maps and symbolication for minified and native stacks
01.3In-app frames separated from vendor noise, with surrounding source
01.4Assign, resolve, ignore, and regression detection when a fixed issue returns

Performance

Transactions and spans give you where the time went, with percentiles computed sampling-aware — a p95 read off a sampled index is weighted rather than counted, so the number means what it says at any sample rate.

Read the docs

02.1Transaction summaries with p50/p95/p99 and throughput
02.2Span waterfalls down to the slow database call
02.3Core Web Vitals per route
02.4Trace-linked errors: jump from a slow request to the exception inside it

Uptime monitoring

A monitor is only down when more than one region agrees it is down. Single-region blips become a degraded state rather than a page, which is the difference between a useful alert and one people learn to ignore.

Read the docs

03.1Multi-region consensus before anything pages you
03.2HTTP, TCP, TLS expiry, DNS, and inbound heartbeat checks
03.3A ninety-day uptime strip that carries state across silent days
03.4Degraded as a real state, distinct from up and from down

On-call and escalation

Incidents drive an escalation policy with repeat and fallback steps. Delivery is confirmed, not assumed: a page that was never delivered escalates rather than sitting quietly as if it had been read.

Read the docs

04.1Rotations, overrides, and coverage gaps shown before they bite
04.2Escalation chains with repeat counts and a fallback rung
04.3Per-user notification rules by urgency, channel, and delay
04.4Push, voice, email, Slack, Teams, Discord, and webhooks

Status pages

Components map to monitors, so the page reflects what your checks actually see instead of what somebody remembered to update. Subscribers are notified on state changes over email or webhook.

Read the docs

05.1Custom domains with managed certificates
05.2Email and webhook subscribers, with unsubscribe handled for you
05.3Maintenance windows that suppress noise without hiding real outages
05.4Incident updates written for the audience that cannot ask you questions

Built for agents

The dashboard and the agent surface are generated from one registry, so a tool can never have different auth or validation than the route it wraps. If you can do it in the UI, your agent can do it too, with the same scoped token.

Read the docs

06.1MCP server with scoped tokens and OAuth 2.1 client registration
06.2Every dashboard page offers the exact call for what is on screen
06.3Docs readable as Markdown, /llms.txt, and over MCP
06.4Mutations audited, so you can see what an agent changed and when

Try it on your own traffic

Point a DSN at us and keep your instrumentation. The free plan needs no card, and the default hard cap means nothing bills you by surprise.

Start free