NEXUS
Production evidence intake and case shaping
Back to incident detail Open command center
Input Channels

Start with the mess. Shape the case before people start relaying it.

Paste raw logs, stack traces, or incident text, watch the evidence get structured, and move into one coordinated triage workflow without the usual handoff overhead.

Supported channels
6
Normalized output
1 envelope
Auth models
Mixed
Launch path
Incident workspace
Demo path
How to read this screen
Bring in noisy evidence, inspect the shaped case, then jump straight into the incident workspace where the crew prepares the final review packet.
1. Paste raw logs

Use the example payload or paste your own incident text.

2. Review parsed evidence

The browser extracts the service, severity, and error signature before the backend reasoning step.

3. Open the console

Jump directly into the incident view where the system turns the shaped evidence into a prepared remediation path for Guardian review.

Live reasoning: ON

Primary Intake Sources

Raw logs are the main proof path. The other sources stay available to show the product can meet teams where incidents actually arrive.

More Sources

Slack, stream anomaly, and batch import remain available without crowding the main demo path.

Raw Incident Intake Preview

This is the moment where unstructured evidence stops being support toil and becomes a case the crew can actually work.
Guided demo bundles

Guided demo bundles

Use a bounded outage bundle when you want a fast, truthful stakeholder walkthrough.

These bundles stay inside the current five-family wedge. They preload realistic evidence without pretending the product supports arbitrary incident classes.
Loading curated bundle choices...
Bundle proof

No demo bundle selected yet

Choose a bundle to explain what the logs should prove, which team the case should route to, and what kind of bounded runtime or debugging evidence to expect.

Expected family
Select a bundle
Likely owner
Select a bundle
Runtime posture
Select a bundle
TRACE posture
Select a bundle
Expected agent path

SENTINEL → PRISM → REPLICA → TRACE → FORGE → GUARDIAN

Choose a bundle to see what the operator should watch for after submit.

Supported incident types

Supported incident families

  • Checkout timeout — Retry amplification causing cascading timeouts
  • DB pool exhaustion — Connection pool saturated, new requests blocked
  • Deploy regression — Recent deployment causing 5xx spike
  • Queue backlog — Worker queue backlogged affecting transaction completion
  • Auth dependency slowdown — Token validation failures from slow auth service

Incidents outside these families will receive a structured error with guidance.

Raw input

This field starts empty so you can paste your own logs immediately. If you want a sample, use Load example logs, then click Submit raw logs.

Parsed evidence

Raw text intake: paste the logs first, then normalize them into the reasoning flow.

Auth: operator paste or copied incident transcript.

The browser will extract the likely service, severity, and error signature before the backend reasoning step.

2026-05-30T10:14:22Z checkout-api ERROR timeout waiting for payment service
2026-05-30T10:14:23Z checkout-api WARN retry budget exhausted
2026-05-30T10:14:25Z payments-worker ERROR queue depth exceeded threshold
service=checkout-api severity=P1
Browser parse
Detected service
checkout-api
Severity hint
P1
Error signature
Timeout / queue pressure
Expected action
Open console
Normalization posture
Awaiting input
Missing signals
Add logs
Hint source
Browser preview
Operator read
Paste logs to shape the case
Live API
Preview only: no raw-log backend endpoint yet
1
Normalize intake
Shape raw logs into a bounded incident packet.
2
Create incident
Persist the fresh `nxs_...` case and preserve intake context.
3
Open workspace
Land at the top incident brief before any deeper review.
Open console
Open incident workspace
No intake parsed yet.

The same incident console will later receive the normalized log bundle and reason over it with the agents.