Incidentary

When an alert fires, the first problem is not fixing the failure — it is getting everyone on the same page about what the failure actually is. War rooms start with five engineers who each have a different fragment of the picture. Incidentary eliminates that first ten minutes.

Incidentary delivers a causal artifact within seconds of an alert. The artifact tells you where the failure first appeared, how it propagated across services, and how much of the causal chain is covered. It is assembled from the traces your services emitted in the window before and after the alert fired. By the time you open the war room, every engineer on the call can read the same artifact.

What Incidentary delivers

Immediate value from one service (Solo Mode)

Install the SDK on a single service and see results in minutes:

• Ghost service detection: Incidentary observes every outbound call your service makes and surfaces uninstrumented dependencies as ghost services — services you depend on but don't have SDK data from. One install reveals your entire dependency topology.
• Service map: A live topology graph showing instrumented services (solid) and ghost services (dashed), with edge health colored by error rate and latency.
• Coverage scorecard: A progress bar showing how much of your service topology is instrumented, with ghost services ranked by call volume so you know where to instrument next.
• Anomaly feed: A zero-configuration feed of detected anomalies — latency spikes, error bursts, retry storms, and dependency degradation — without waiting for an incident or configuring alerts.
• Request waterfall: A per-request view of every outbound call from one inbound request, with pattern detection for sequential calls, N+1 queries, retry storms, and outlier latency.

Team value during incidents

When an incident is triggered, you get:

• Where the failure first broke: the earliest event in the causal chain with a non-success status, identified by service name, operation type, and status code
• How it propagated: the full causal chain from the root request through every instrumented service
• How much is covered: a completeness label (full, partial, or low) that tells you whether any links in the chain are missing, and where
• A shared URL: a publicly accessible link to the incident timeline you can drop into Slack, PagerDuty, or a war room thread without requiring authentication

What Incidentary is not

• Not a metrics platform: The anomaly feed surfaces detected patterns, but Incidentary is not a time-series dashboard. For metric exploration, use Datadog, Grafana, or equivalent.
• Not a log aggregator: Incidentary captures structured causal events, not free-form log lines.
• Not an AI diagnostics tool: Incidentary makes no speculative claims about what caused a failure. It shows what happened, as recorded. Everything in the artifact is a fact, not an inference.
• Not an alerting system: The anomaly feed is informational, not interruptive. Anomalies appear in the feed — they do not page you. Incidentary integrates with your existing alerting tools (PagerDuty, OpsGenie) to trigger incident assembly.
• Not a replacement for your existing observability stack: Incidentary is a precursor to your existing tools. You read Incidentary first, then go to Datadog knowing exactly what you are looking for.

How engineers use it

Day 1 (solo): Install the SDK on one service. Within minutes, see ghost services, a request waterfall, and the anomaly feed. You get value before anyone else on your team touches it.

Week 1 (expansion): Share your first incident trace in Slack. Teammates see the causal chain, notice ghost service gaps, and instrument their own services. The coverage scorecard tracks progress.

Ongoing (team convergence): When an incident starts, every engineer reads the same artifact. Incidentary compresses the alignment phase so the investigation starts from a shared, verified starting point instead of from individual hunches.

Concepts

• The Assembled Artifact — what the incident artifact contains and how to read it
• Pre-Alert Capture — why artifacts are available within seconds
• Ghost Services — uninstrumented dependencies detected from outbound calls
• Service Map — live topology of your service dependencies
• Anomaly Feed — zero-config detection of latency spikes, error bursts, and more

Get started

Install the SDK for your language:

• Node.js
• Python
• Go

Already using OpenTelemetry? Send traces to Incidentary with zero code changes:

• OpenTelemetry — point your Collector or SDK exporter at Incidentary's OTLP endpoint

Connect your alert source:

• Slack
• PagerDuty
• OpsGenie