Skip to content

Log Center

The Log Center is used to view Nekro Agent's operational logs within the WebUI. It is useful for troubleshooting issues related to deployment, adapters, model calls, plugins, workspaces, and sandboxes.

Logs May Contain Sensitive Information

When troubleshooting, be aware that logs may contain channel names, user information, request URLs, error context, and similar data. Before seeking help publicly, it is recommended to redact sensitive details.

What the Log Center Can Do

FeatureDescription
View Historical LogsReview recent operational records and error messages
Filter by SourceView output from a specific module or service only
Real-time Log StreamContinuously observe new logs while the page is open
Download LogsSave log files for feedback or archiving
Filter by Severity LevelLocate issues based on INFO, WARNING, ERROR, and other levels

When to Check Logs

Open the Log Center in the following situations:

  • The bot is not replying, but the adapter shows it is online
  • Model calls are failing, timing out, or returning errors
  • Plugin commands fail to execute
  • Workspace or Claude Code sandbox fails to start
  • Knowledge base upload, vectorization, or retrieval is abnormal
  • The WebUI page displays an unknown error

Understanding Log Levels

LevelMeaningRecommendation
DEBUGDetailed debugging informationEnable temporarily for troubleshooting only
INFONormal operational informationMonitor during routine observation
WARNINGIssues that may affect functionalityCheck configuration if they appear frequently
ERRORFeature execution failurePrioritize resolution
CRITICALSevere anomalyInvestigate service status as soon as possible

For more detailed logs, adjust APP_LOG_LEVEL in the "System Configuration" page. After troubleshooting, it is recommended to restore the default level to avoid excessive log output.

Troubleshooting Suggestions

Adapter Connection Issues

First filter or search for the adapter name, such as OneBot, Telegram, Discord, Feishu, WeCom, etc. Focus on:

  • Authentication failures
  • Incorrect webhook URLs
  • Network connection failures
  • Message parsing errors

Model Call Issues

Search for the model group name, Base URL, or error keywords. Common causes include:

  • Incorrect API Key or insufficient quota
  • Incompatible Base URL
  • Incorrectly entered model name
  • Request timeouts
  • Upstream service rate limiting

Workspace and Sandbox Issues

If a workspace task fails, first check the workspace page for hints, then search the Log Center for keywords such as "workspace," "sandbox," or "Claude Code."

Common causes include image pull failures, insufficient port ranges, sandbox startup timeouts, and abnormal workspace paths.

Plugin Issues

Plugin-related errors typically include the plugin name or command name. When troubleshooting, try temporarily disabling recently added or updated plugins to narrow down the scope.

Downloading Logs for Feedback

When reporting issues to the community or developers, it is recommended to provide:

  1. The approximate time the issue occurred
  2. The relevant log excerpts
  3. The adapter and deployment method used
  4. Whether any recent version updates or configuration changes were made

Please remove sensitive information such as API keys, tokens, cookies, and internal network addresses before sending.

Log-related configuration can be found in Advanced Configuration. Common settings include:

  • APP_LOG_LEVEL
  • SAVE_PROMPTS_LOG
  • DEBUG_IN_CHAT