Traceability

Traceability provides complete visibility into agent and workflow executions, enabling debugging, auditing, and compliance.

What is Traceability?

Every execution in FlowGenX generates a comprehensive trace that captures:

Input/Output - Request and response data
Decision Path - Agent reasoning and choices
Tool Calls - External API invocations
Timing - Duration of each step
Errors - Failures and exceptions

Trace Components

Execution Trace

The top-level trace for a workflow or agent invocation.

Trace ID: tr_abc123
Status: Completed
Duration: 2.3s
Started: 2024-01-15 10:30:00

Spans

Individual operations within the trace:

Agent reasoning spans
Tool execution spans
LLM call spans
Data transformation spans

Events

Point-in-time occurrences:

State changes
Condition evaluations
Error events
Custom annotations

Using the Trace Viewer

Accessing Traces

Navigate to Operations > Traces
Filter by time range, status, or agent
Click on a trace to view details

Trace Timeline

Visual representation of execution flow:

Waterfall view of spans
Parallel execution visibility
Critical path highlighting

Span Details

Drill into any span to see:

Input parameters
Output results
Duration breakdown
Associated logs

Search and Filter

By Trace ID - Direct lookup
By Time Range - Filter by execution time
By Status - Success, failure, timeout
By Agent/Workflow - Filter by resource
By User - Filter by initiating user
Full-text Search - Search in request/response data

Retention and Export

Retention Policies

Configure how long traces are retained:

Development: 7 days
Staging: 30 days
Production: 90 days (configurable)

Export Options

JSON export for analysis
Integration with external observability tools
Compliance report generation

Best Practices

Add Custom Annotations - Tag important business events
Monitor Error Traces - Set up alerts for failures
Sample High-Volume Traces - Use sampling for cost control
Correlate with Metrics - Link traces to dashboard metrics