Debug Command Guide
This document introduces the debugging and system maintenance commands of Nekro Agent. These commands are mainly used for advanced operations such as technical debugging, performance monitoring, and system maintenance.
Risk Warning
The commands in this document have high system privileges and may affect system stability or data security. Please ensure you use them cautiously while understanding the function of the commands.
Platform Limitation
Currently, all commands are only available under the OneBot V11 adapter (i.e., QQ platform), and other protocol endpoints are not yet supported.
Permission Description
All debug commands require super administrator privileges (configured in config.SUPER_USERS).
Code Execution and Debugging
exec
Function: Execute Python code in the sandbox and return results
Syntax: exec <python_code>
Permission Requirement: Super administrator
Parameter Description:
python_code: Python code to be executed
Security Description:
- Code is executed in a sandbox environment
- Subject to execution time and resource limitations
- Has certain system resource access permissions
Usage Examples:
exec print("Hello World")
exec import os; print(os.getcwd())
exec 1 + 1code_log / code-log
Function: View code execution history
Syntax: code_log [index]
Permission Requirement: Super administrator
Parameter Description:
index(optional): Record index number- Positive number: View the Nth record in chronological order
- Negative number: View the Nth record in reverse chronological order
- Omit: View the latest record
Displayed Content:
- Executed code content
- Execution result/output
- Execution time information
Usage Examples:
code_log # View the latest execution record
code_log -2 # View the 2nd to last record
code_log 1 # View the 1st recordSystem Messages and Debug Mode
system
Function: Add a system message to the current chat and trigger the Agent
Syntax: system <message>
Permission Requirement: Super administrator
Parameter Description:
message: System message content to be added
Usage Scenarios:
- Debug Agent behavior
- Inject specific system instructions
- Test prompt responses
Usage Examples:
system Please introduce yourself
system Current task: Analyze the data provided by the userdebug_on / debug-on
Function: Enable prompt debug mode
Syntax: debug_on
Permission Requirement: Super administrator
Function Description:
- Exit role-playing mode
- Focus on technical analysis and status reporting
- Provide additional debug information after each response
- Analyze current context status and settings
Debug Mode Features:
- Technical analysis priority
- Display prompt strengths/weaknesses
- Identify potential issues
- Provide improvement suggestions
Usage Examples:
debug_ondebug_off / debug-off
Function: Disable prompt debug mode
Syntax: debug_off
Permission Requirement: Super administrator
Function Description:
- Return to normal role-playing mode
- Stop debug analysis output
- Ignore all debug contexts
Usage Examples:
debug_offConfiguration Management
conf_show / conf-show
Function: View or list system configuration
Syntax: conf_show [config_key]
Permission Requirement: Super administrator
Parameter Description:
config_key(optional): Name of the configuration item to view- When omitted: Display list of all dynamically modifiable configuration items
Supported Configuration Types:
- Integer (int)
- Float (float)
- Boolean (bool)
- String (str)
Usage Examples:
conf_show # List all modifiable configuration items
conf_show AI_GENERATE_TIMEOUT # View AI generation timeout configurationconf_set / conf-set
Function: Dynamically modify system configuration
Syntax: conf_set <key>=<value>
Permission Requirement: Super administrator
Parameter Description:
key: Configuration item namevalue: New configuration value
Data Type Handling:
- Integer/Float: Automatic type conversion
- Boolean: Supports
true/false,1/0,yes/no - String: Direct assignment
Usage Examples:
conf_set AI_GENERATE_TIMEOUT=60
conf_set ENABLE_DEBUG_MODE=true
conf_set AI_COMMAND_OUTPUT_PREFIX=[Bot]conf_reload / conf-reload
Function: Reload configuration file
Syntax: conf_reload
Permission Requirement: Super administrator
Function Description:
- Reread all settings from the configuration file
- Discard dynamically modified configurations
- Restore to the configuration state in the file
Usage Examples:
conf_reloadconf_save / conf-save
Function: Save current configuration to file
Syntax: conf_save
Permission Requirement: Super administrator
Function Description:
- Write configurations in memory to the configuration file
- Persist dynamically modified settings
Usage Examples:
conf_saveModel Testing
model_test / model-test
Function: Test model availability and performance
Syntax: model_test [options] <model_names...>
Permission Requirement: Super administrator
Option Parameters:
-g: Treat parameters as model group names rather than model names--stream: Use streaming mode for testing--use-system: Include system messages in testing
Wildcard Support:
- Supports
*wildcard to match multiple models/model groups - For example:
gpt*matches all models starting with "gpt"
Test Content:
- Model response capability
- Response speed statistics
- Success/failure rate
- Performance benchmarking
Usage Examples:
model_test gpt-4o # Test a single model
model_test -g main_model debug_model # Test model groups
model_test --stream gpt* # Stream test all gpt models
model_test --use-system claude* # Include system message testingError Log Management
log_err_list / log-err-list / log_err_ls / log-err-ls
Function: View error log list
Syntax: log_err_list [options]
Permission Requirement: Super administrator
Option Parameters:
-p <page_number>: Specify page number (default: 1)-s <per_page>: Number of items per page (default: 10, max: 50)-a/--all: View all files in the log directory instead of cached list
Displayed Content:
- Error log file list
- File modification time
- Log number (used for reference by other commands)
Usage Examples:
log_err_list # View recent error logs
log_err_list -p 2 -s 5 # View page 2, 5 items per page
log_err_list -a # View all log fileslog_chat_test / log-chat-test
Function: Test LLM requests using conversation content from error logs
Syntax: log_chat_test <log_index_or_filename> [options]
Permission Requirement: Super administrator
Parameter Description:
log_index_or_filename: Log index number or filename
Option Parameters:
-g <model_group_name>: Specify the model group to use (default: main model group)--stream/-s: Use streaming mode
Log Matching Rules:
- Try to parse as an index number
- Match by filename in the cached list
- Directly search for files in the log directory
- Automatically add
.jsonsuffix and retry
Test Content:
- Reproduce requests from error logs
- Test model response capability
- Performance benchmarking
- Error reproduction analysis
Usage Examples:
log_chat_test 1 # Test the 1st log
log_chat_test error_20241201.json # Test specified log file
log_chat_test 2 -g debug_model # Use debug model group for testing
log_chat_test 3 --stream # Stream mode testingSystem Maintenance
clear_sandbox_cache / clear-sandbox-cache / na_csc / na-csc
Function: Clear the cache directory of the sandbox environment
Syntax: clear_sandbox_cache
Permission Requirement: Super administrator
Cleared Content:
- pip cache directory
- Sandbox package directory
- Temporary files and directories
Displayed Information:
- Number of files cleared
- Freed disk space
- Clear operation status
Usage Examples:
clear_sandbox_cachegithub_stars_check / github-stars-check
Function: Check if the user has starred the official GitHub repository
Syntax: github_stars_check
Permission Requirement: Super administrator
Displayed Content:
- Star status overview
- List of starred repositories
- List of unstarred repositories
Usage Examples:
github_stars_checkHigh-Risk Commands
Dangerous Operations
The following commands have extremely high system risks and may cause data loss or system instability. Please use them with full understanding of the consequences.
docker_restart / docker-restart
Function: Restart Docker container
Syntax: docker_restart [container_name]
Permission Requirement: Super administrator + Docker environment
Parameter Description:
container_name(optional): Container name (default: nekro_agent)
Limitations:
- Only available in Docker environment
- Requires Docker operation permissions
Usage Examples:
docker_restart # Restart default container
docker_restart my_container # Restart specified containerdocker_logs / docker-logs
Function: View Docker container logs
Syntax: docker_logs [container_name]
Permission Requirement: Super administrator + Docker environment
Parameter Description:
container_name(optional): Container name (default: nekro_agent)
Displayed Content:
- Last 100 lines of container logs
- Running status information
Usage Examples:
docker_logs # View default container logs
docker_logs my_container # View specified container logssh
Function: Execute Shell commands
Syntax: sh <command>
Permission Requirement: Super administrator
Parameter Description:
command: Shell command to be executed
Security Risks:
- Has complete system access permissions
- May affect system stability
- Command execution results are irreversible
Usage Examples:
sh ls -la # List files
sh df -h # View disk usage
sh ps aux | grep python # View Python processesnekro_db_reset / nekro-db-reset
Function: Reset database
Syntax: nekro_db_reset [-y] [table_name]
Permission Requirement: Super administrator
Parameter Description:
-y: Confirm to execute reset operationtable_name(optional): Specify the data table to reset
Security Mechanism:
- Requires two executions within 1 minute for confirmation
- First execution only displays warning
- Second execution requires
-yparameter for confirmation
Data Impact:
- Completely clear specified data table or entire database
- Operation is irreversible
- Will lose all related data
Usage Examples:
nekro_db_reset # First execution, displays warning
nekro_db_reset -y # Confirm to reset entire database
nekro_db_reset -y chat_messages # Reset specified data tableSecurity Recommendations
- Permission Control: Ensure only trusted users have super administrator privileges
- Operation Records: It is recommended to back up relevant data before important operations
- Test Environment: Test high-risk commands outside the production environment first
- Log Monitoring: Regularly check command execution logs to monitor abnormal operations
- Configuration Backup: Use
conf_saveto backup before important configuration modifications
Getting Help
- Use the
na_helpcommand to view command overview - View Basic Command Guide to understand daily management functions
- Visit GitHub Project to report issues
