Shell Executor Skill

Execute shell commands through your Moltbot agent.

Shell Executor

Execute shell commands through your AI agent. This is a powerful but potentially dangerous skill.

High Risk Skill: This skill can execute arbitrary commands on your system. Only enable if you fully understand the security implications.

What It Does

Run bash/shell commands
Execute scripts
Manage system processes
Automate system administration

Install / Enable

# .env
SKILL_SHELL_ENABLED=true
SKILL_SHELL_ALLOWED_COMMANDS=ls,cat,echo,pwd  # Optional whitelist
SKILL_SHELL_WORKING_DIR=/home/user/safe-dir    # Optional sandbox

Permissions Required

Permission	Description
System	Full shell access (or whitelisted commands)
Network	If running network commands
Files	Depends on commands executed

Example Commands

User: List files in the current directory
Agent: [Executes: ls -la]
       drwxr-xr-x  5 user user  160 Jan 15 10:00 .
       drwxr-xr-x  3 user user   96 Jan 15 09:00 ..
       -rw-r--r--  1 user user 1234 Jan 15 10:00 file.txt

User: How much disk space is available?
Agent: [Executes: df -h]
       Filesystem      Size  Used Avail Use% Mounted on
       /dev/sda1       50G   20G   28G  42% /

User: What's the server uptime?
Agent: [Executes: uptime]
       10:30:00 up 45 days, 3:22, 1 user, load average: 0.00, 0.01, 0.00

Configuration

Variable	Description
`SKILL_SHELL_ENABLED`	Set to `true` to enable
`SKILL_SHELL_ALLOWED_COMMANDS`	Whitelist of allowed commands
`SKILL_SHELL_BLOCKED_COMMANDS`	Blacklist of blocked commands
`SKILL_SHELL_WORKING_DIR`	Restrict to this directory
`SKILL_SHELL_TIMEOUT`	Command timeout in seconds
`SKILL_SHELL_MAX_OUTPUT`	Max output characters

Security Configuration

Recommended: Whitelist approach

# Only allow specific safe commands
SKILL_SHELL_ALLOWED_COMMANDS=ls,cat,head,tail,grep,wc,date,uptime,df,free,ps

Alternative: Blacklist approach (less secure)

# Block dangerous commands
SKILL_SHELL_BLOCKED_COMMANDS=rm,dd,mkfs,fdisk,shutdown,reboot,sudo,su,chmod,chown

Sandboxing

# Restrict to specific directory
SKILL_SHELL_WORKING_DIR=/home/user/sandbox

# Combined with command whitelist
SKILL_SHELL_ALLOWED_COMMANDS=ls,cat,head,tail

Risks

System Damage: Commands like rm -rf / can destroy your system.

Data Exfiltration: Malicious prompts could leak sensitive files.

Privilege Escalation: If running as root, full system compromise is possible.

Resource Exhaustion: Fork bombs or infinite loops can crash the system.

Best Practices

Never run as root - Use a limited user account
Use command whitelists - Only allow necessary commands
Set working directory - Sandbox to a safe location
Enable audit logging - Log all executed commands
Set timeouts - Prevent long-running commands
Review prompts - Understand how AI interprets requests

Troubleshooting

Command not found

# Check if command is in PATH inside container
docker exec moltbot which ls

# Or add to PATH
docker exec moltbot echo $PATH

Permission denied

The container user lacks permissions. See Permission Denied.

Output truncated

Increase SKILL_SHELL_MAX_OUTPUT or pipe to head:

SKILL_SHELL_MAX_OUTPUT=10000

Alternatives

Code Interpreter - Safer Python execution
Git Operations - Specific git commands only