Cognition: AI Analysis and Reports
The Skytells AI engine audits your Cognition event data and generates a structured report covering errors, threats, runtime health, and recommended actions. Available in Console under Cognition / AI Analysis.
AI Analysis and Reports require a Pro plan or higher. See Plans for a tier comparison, or Upgrade to Pro ↗ directly.
When Cognition sends events to Skytells, the AI engine audits them and builds a structured report. This isn't a raw log dump — it's a human-readable assessment that groups patterns, identifies correlations between errors and runtime conditions, and lists specific recommended actions.
AI Analysis Tab
Console: Cognition Console ↗ → AI Analysis tab
The AI Analysis tab is where the generated report lives. It's organised into sections:
| Section | What the AI covers |
|---|---|
| Health Summary | A written assessment of your application's current state across all event types |
| Error Analysis | Grouped error patterns, recurrence trends, affected users, and root-cause hypotheses |
| Security Audit | Threat classification, attack pattern recognition, and suggested mitigations |
| Runtime Assessment | Anomaly correlation — for example, heap growth before an error spike, or ELU increases under traffic load |
| Recommendations | Prioritised, actionable next steps based on the full event picture |
| Historical Comparison | How current metrics compare to the previous reporting period |
The report updates automatically as new events arrive. There is no manual refresh.
Configuring the AI Engine
The AI engine is configured per project in the Console:
- Open the Cognition Console ↗
- Go to the AI Analysis tab
- Open AI Settings
From there you can control:
- What it monitors — choose which event types feed the analysis
- Thresholds — define the conditions that trigger escalation or action
- Actions — what the AI does when a condition is met: report, alert, or auto-remediate
- Reporting cadence — how often the AI refreshes its assessment
AI configuration is project-scoped. Different projects can have different AI behaviors, thresholds, and actions configured independently.
Live Events
Console: Cognition Console ↗ → Live tab
The Live tab streams events from your application in real time. You'll see errors, runtime snapshots, security threats, anomalies, and trace spans as they arrive, without refreshing the page.
This is useful for:
- Verifying the SDK — confirm events are coming through right after a deployment
- Watching a rollout — keep an eye on error rates and anomalies while a new version goes out
- Tracking an active incident — see the exact sequence of events as they happen
Live events aren't paginated history. For historical queries, use the Errors, Security Threats, or Runtime Health tabs.
How the Report Is Generated
- The SDK batches and ships events to
dsn.skytells.aivia the transport pipeline - The ingestion layer fans events to both the Live stream and the AI engine
- The AI engine audits the incoming data against your project's history and configuration
- A structured report is written or updated in the AI Analysis tab
Related
- Analytics — Full Console tab reference and CLI commands
- Configuration —
beforeSend, runtime thresholds - Transport — How events are delivered to
dsn.skytells.ai - Plans — Pro plan feature comparison
How is this guide?
Analytics
How to read your Cognition data in the Skytells Console and CLI. Covers all Console views, every CLI command, and common investigation workflows.
Examples
Working integration patterns for Express, Fastify, Koa, background workers, microservices, PII scrubbing, OpenTelemetry, health endpoints, and more.