Logs

The logs panel provides detailed records of all AI agent executions, showing how requests were processed, which nodes executed, and where errors occurred.

Overview

Logs provide:

• Debugging: Quickly identify problems and errors
• Monitoring: Track agent performance
• Analysis: Understand how the agent makes decisions
• Optimization: Identify bottlenecks in logic

When to Use Logs

• After chat testing — verify execution details
• When errors occur — find the cause
• To analyze tool usage
• To verify data correctness between nodes

Logs Panel

Opening the Panel

The logs panel is located at the bottom of the agent editor:

1. Open your AI agent in the editor
2. The logs panel displays automatically at the bottom
3. If closed, click the Logs button in the bottom of the interface

Panel Structure

The logs panel has three sections:

Resizing Panels

• Panel height: Drag the top border up/down
• Section width: Drag dividers between sections left/right

Execution History

The left section shows all agent executions.

Execution List

Each execution displays:

• Time: When it was run (HH:mm • DD MMM YYYY)
• Status: Execution state icon
• Duration: How long execution took (milliseconds)

Execution statuses:

Selecting an Execution

1. Click on an execution in the list
2. The execution is highlighted
3. The Nodes section loads executed nodes
4. Executed nodes are highlighted on the canvas

Filtering

Filters help find specific executions among many logs.

Time Filter

Available options:

• Last hour: Executions from the past hour
• Last day: Executions from the past 24 hours
• Last week: Executions from the past 7 days
• Manually: Specify exact date range

Executor Filter

Filter by how the agent was triggered:

• Manual: Runs from chat via the Run button
• API: Calls through API endpoints
• Scheduled: Automatic runs via cron

Parent Filter

• Has Parent: Show only executions called from other agents
• No Parent: Show only root executions

Sorting and Refresh

• Sort: Toggle between newest first (default) or oldest first
• Refresh: Update the logs list with new executions

Viewing Nodes

The middle section shows nodes executed in the selected run.

Node List

Displays:

• Execution order: Nodes listed in order of execution
• Node name: Name from the canvas
• Node type: Start, AI Agent, Tool, Output, etc.
• Status: Successfully executed or failed with error

Selecting a Node

1. Click on a node in the list
2. The canvas automatically scrolls to that node
3. The Details section shows execution details

Execution Details

The right section shows detailed information about the selected node.

Detail Structure

Input:

• Data the node received
• Format: JSON
• Expandable structure

Output:

• Result of node execution
• What the node passed to subsequent nodes
• Format: JSON

Execution Info:

• Start time
• End time
• Duration
• Execution ID

Error:

• If the node failed, displays:
  • Error message
  • Stack trace (if available)
  • Error code

Debugging

Logs are the primary tool for debugging agent issues.

Finding Errors

Step 1: Find the problematic execution

1. Open the logs panel
2. Find an execution with Failed status
3. Click on it

Step 2: Identify the problematic node

1. In the Nodes section, find the node with error (usually marked red)
2. Click on it

Step 3: Study the error details

1. In Details, scroll to the Error section
2. Read the error message
3. Analyze the input data

Common Debugging Scenarios

Agent doesn't call the expected tool

1. Find the execution in logs
2. Check the node list — is the Tool there?
3. If Tool is missing:
   • Verify the Tool is connected to the AI Agent node
   • Check Tool description (agent uses it for selection)
   • Improve agent prompt to be more explicit

Node outputs incorrect data

1. Find the node in logs
2. Check Input — is input data correct?
3. Check Output — what exactly is incorrect?
4. Trace back through the chain to find where data went wrong

External service connection error

1. Find the error in Details → Error
2. Common errors:
   • 401 Unauthorized → Check API key in connector
   • 403 Forbidden → Check access permissions
   • 404 Not Found → Check URL/endpoint
   • Timeout → Check network connection and timeouts

Performance Analysis

• Each execution shows total time (ms)
• Each node shows its execution time
• Find nodes with high duration — these are bottlenecks

Optimization tips:

• Slow Tool → optimize the request or increase timeout
• Slow AI Agent → reduce max tokens or simplify prompt
• Slow JS node → optimize the code

Best Practices

Regular Monitoring

• Check logs after each deployment
• Monitor errors in production
• Analyze performance weekly

Using Filters

• Time filter for analyzing specific periods
• Executor filter for separating dev/production analysis
• Save problematic executions for analysis

Debugging

• Always start with Failed executions
• Check Input for each node
• Compare successful and problematic executions
• Document found issues

Troubleshooting

Logs don't load

• Refresh the page
• Click the Refresh button in the logs panel
• Check filters — they may be too restrictive

New executions not showing

• Click Refresh
• Check time filter — increase the period
• Verify the agent actually executed

Node details not visible

• Ensure an execution is selected in History
• Click on a node in the Nodes section
• Refresh the page if details don't load