Skip to main content
Untrace Dashboard Light

​
Overview

The Untrace Dashboard provides a powerful web interface for monitoring your LLM traces, configuring routing rules, analyzing performance metrics, and managing your observability integrations. Access it at https://untrace.dev/app.

Real-time Monitoring

View LLM traces as they flow through your system

Routing Configuration

Set up intelligent routing rules for your traces

Analytics

Analyze costs, performance, and usage patterns

Integrations

Configure connections to observability platforms

​
Getting Started

​
Accessing the Dashboard

  1. Navigate to https://untrace.dev/app
  2. Sign in with your Untrace account
  3. You’ll land on the overview page showing your LLM trace activity

​
Dashboard Layout

The dashboard is organized into several key sections:
  • Navigation Sidebar: Quick access to all dashboard features
  • Main Content Area: Displays your selected view (traces, analytics, settings, etc.)
  • Activity Feed: Real-time trace activity stream
  • Status Bar: Connection status and platform health indicators

​
Real-time Monitoring

​
Live Trace Feed

Monitor LLM traces as they flow through Untrace:
View traces in a chronological list with key information:
  • Timestamp
  • Model (GPT-4, Claude, etc.)
  • Provider (OpenAI, Anthropic, etc.)
  • Token usage
  • Cost
  • Latency
  • Routing destinations
  • Status (success/error)
Filter the trace feed to focus on what matters: 
// Example filters
{
  model: "gpt-4",
  provider: "openai",
  status: "success",
  costRange: { min: 0.01, max: 0.10 },
  timeRange: "last-hour"
}
Available filter options:
  • Model: Filter by specific models (GPT-4, Claude-3, etc.)
  • Provider: OpenAI, Anthropic, Google, etc.
  • Status: Success, Failed, Rate Limited
  • Cost Range: Filter by token cost
  • Latency: Response time thresholds
  • Destinations: Filter by routing destinations
  • Time Range: Last hour, 24 hours, 7 days, custom range
  • Tags: Custom tags and metadata

​
Routing Configuration

​
Creating Routing Rules

Configure how traces are routed to different platforms:
1

Create Rule

Click New Routing Rule and configure:
  • Rule name and description
  • Matching conditions
  • Destination platforms
  • Priority order
2

Define Conditions

Set up matching conditions:
  • Model type (GPT-4, Claude, etc.)
  • Cost thresholds
  • Error conditions
  • Custom metadata
  • Environment tags
3

Select Destinations

Choose where to send matching traces:
  • Primary destination
  • Fallback destinations
  • Multi-destination routing
  • Sampling rates
4

Test Rule

Test your routing rule:
  • Send test traces
  • Verify routing behavior
  • Check destination delivery

​
Routing Examples

Common routing patterns: 
name: "Route GPT-4 to LangSmith"
conditions:
  model: "gpt-4*"
destinations:
  - platform: "langsmith"
    sample_rate: 1.0

name: "High-cost trace analysis"
conditions:
  cost: "> 0.10"
destinations:
  - platform: "langfuse"
    tags: ["high-cost", "analyze"]

name: "Failed request debugging"
conditions:
  status: "error"
destinations:
  - platform: "keywords-ai"
  - platform: "custom-webhook"
    url: "https://api.yourapp.com/errors"

name: "Platform comparison"
conditions:
  model: "claude-3-opus"
destinations:
  - platform: "langsmith"
    sample_rate: 0.5
  - platform: "langfuse"
    sample_rate: 0.5

​
Analytics

​
Overview Metrics

View key metrics at a glance:
  • Total Traces: Daily, weekly, monthly counts
  • Token Usage: Total tokens processed
  • Total Cost: Aggregate costs across all models
  • Average Latency: P50, P95, P99 response times
  • Error Rate: Failed requests and error types
  • Model Distribution: Usage by model type

​
Cost Analysis

Deep dive into your LLM costs:
  • Cost breakdown by model
  • Token usage per model
  • Average cost per request
  • Cost trends over time

​
Performance Metrics

Monitor LLM performance:
  • Latency Distribution: Response time histograms
  • Token/Second: Throughput metrics
  • Queue Depth: Pending requests
  • Error Analysis: Error types and frequencies
  • Rate Limit Tracking: Provider limit utilization

​
Custom Reports

Generate custom analytics reports:
  1. Select metrics to include
  2. Choose aggregation period
  3. Apply filters
  4. Export as CSV or PDF
  5. Schedule automated reports

​
Integrations

​
Managing Platform Connections

Configure connections to observability platforms:
1

Add Integration

Click New Integration and select platform:
  • LangSmith
  • Langfuse
  • Keywords.ai
  • Helicone
  • Custom webhook
2

Configure Authentication

Provide platform credentials:
  • API keys
  • OAuth tokens
  • Webhook URLs
  • Custom headers
3

Set Defaults

Configure default settings:
  • Default tags
  • Metadata mapping
  • Retry policies
  • Timeout settings
4

Test Connection

Verify the integration:
  • Send test trace
  • Check delivery status
  • Verify data format

​
Platform-Specific Settings

Configure platform-specific features:

​
LangSmith

  • Project mapping
  • Environment tags
  • Custom metadata fields
  • Feedback integration

​
Langfuse

  • Session tracking
  • User identification
  • Score mappings
  • Public link generation

​
Keywords.ai

  • Cost tracking settings
  • Alert thresholds
  • Custom dashboards
  • API quota management

​
Custom Webhooks

  • Payload transformation
  • Authentication headers
  • Retry configuration
  • Response validation

​
Team Management

​
Access Control

Manage team access and permissions:

​
API Keys

Manage API keys for different environments: 
# Production key with full access
UNTRACE_API_KEY=utr_prod_xxx

# Development key with limited access
UNTRACE_API_KEY=utr_dev_xxx

# CI/CD key for automated testing
UNTRACE_API_KEY=utr_ci_xxx

​
Advanced Features

​
Trace Sampling

Configure intelligent sampling to reduce costs: 
// Sampling configuration
{
  "default_sample_rate": 0.1,  // 10% default
  "rules": [
    {
      "condition": "model == 'gpt-4'",
      "sample_rate": 0.05  // 5% for expensive models
    },
    {
      "condition": "error == true",
      "sample_rate": 1.0  // 100% for errors
    },
    {
      "condition": "cost > 0.50",
      "sample_rate": 1.0  // 100% for high-cost requests
    }
  ]
}

​
PII Detection

Configure privacy protection:
  • Automatic Detection: Identify potential PII
  • Redaction Rules: Define what to redact
  • Allowlist: Specify safe patterns
  • Audit Trail: Track redaction events

​
Alerting

Set up alerts for important events:
  • Daily spend thresholds
  • Unusual cost spikes
  • Budget warnings
  • Model-specific limits

​
Troubleshooting

​
Common Issues

  • Verify your API key is correct
  • Check network connectivity
  • Ensure proper SDK initialization
  • Verify routing rules are active
  • Check platform credentials
  • Verify network access
  • Review error logs
  • Test with minimal payload
  • Check routing rule complexity
  • Review sampling configuration
  • Monitor platform status
  • Consider regional deployment

​
Debug Mode

Enable debug mode for detailed diagnostics:
  1. Go to Settings → Advanced
  2. Toggle Debug Mode
  3. View detailed trace logs
  4. Export diagnostic bundle

​
API Access

Access dashboard functionality programmatically: 
# Get trace history
curl -X GET https://untrace.dev/api/v1/traces \
  -H "Authorization: Bearer YOUR_API_KEY"

# Get analytics data
curl -X GET https://untrace.dev/api/v1/analytics \
  -H "Authorization: Bearer YOUR_API_KEY"

# Update routing rules
curl -X PUT https://untrace.dev/api/v1/routing/rules \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d @routing-rules.json
See the API Reference for complete documentation.

​
Next Steps

SDK Integration

Integrate Untrace SDK in your applications

Routing Guide

Learn advanced routing strategies

Provider Setup

Configure LLM provider connections

API Reference

Explore the complete API documentation