A pre-flight security validation skill that scans shell commands for dangerous patterns, destructive operations, and data leaks before they execute.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install exec-security
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install exec-security using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Exec Security?
Exec Security serves as a critical defensive layer for AI agents and developers interacting with terminal environments. It provides a comprehensive pre-execution audit for shell commands, identifying high-risk patterns such as recursive deletions, unauthorized system file modifications, and credential exfiltration. By integrating this into your Openclaw Skills workflow, you ensure that programmatic command generation remains within safe boundaries.
The skill categorizes risks into four severity levels, ranging from critical blocks for destructive actions like filesystem wipes to warnings for suspicious Unicode obfuscation or resource exhaustion. It doesn't just block; it educates by suggesting safe alternatives like using length checks for secrets instead of echoing them, or utilizing recovery-focused tools like trash instead of permanent removal.
Exec Security Use Cases
- Validating commands constructed programmatically from external data or LLM outputs.
- Protecting sensitive directories such as .ssh or /etc/ from accidental or malicious modification.
- Preventing accidental credential exposure in logs or terminal output during automated scripts.
- Auditing complex piped commands and remote scripts (download-and-execute) before they touch the local system.
- Detecting hidden Unicode attacks or homoglyphs in commands sourced from untrusted web content.
How Exec Security Works
- The skill intercepts the proposed shell command string before it is passed to the system executive.
- It performs a multi-stage scan looking for destructive patterns, sensitive path references, and shell-specific obfuscation techniques.
- The command is mapped against a risk matrix to determine the security level: Critical, High, Medium, or Low.
- If a risk is detected, the skill provides a detailed explanation of the threat and, where possible, a safer command alternative.
- Depending on the risk level, the skill either blocks execution entirely, requests explicit user confirmation, or issues a warning before proceeding.
Exec Security Setup
To integrate this security check into your environment, add the logic to your Openclaw Skills configuration. While it is advisory, it works best when paired with native execution approval systems.
# Example of checking a command before execution
# 1. Pass the command to the exec-security scanner
# 2. Review the risk level and suggested alternatives
# 3. Proceed only if risk is Low or user-confirmed
Exec Security Data Schema & Taxonomy
Exec Security utilizes a structured risk taxonomy to organize its audit results. The following table describes the primary risk categories used within the Openclaw Skills framework:
| Category | Description | Examples |
|---|---|---|
| Destructive | Permanent data loss operations | rm -rf, dd, mkfs |
| Exfiltration | Outbound data or secret leaks | curl -d @file, echo $API_KEY |
| System Integrity | Unauthorized system file changes | /etc/passwd, .ssh/authorized_keys |
| Injection | Shell manipulation or RCE | eval, LD_PRELOAD, curl |
| Resources | Denial of service patterns | fork bombs, disk fill, infinite loops |
Exec Security Advanced Features
- Multi-shell support covering specific vulnerabilities in both Bash (hex-encoding, namerefs) and Zsh (zmodload, equals expansion).
- Automated path resolution using readlink to prevent Symlink and Time-of-Check to Time-of-Use (TOCTOU) attacks.
- Context-aware variable scanning that distinguishes between generic environment variables and sensitive credentials.
- Heuristic detection of Unicode obfuscation, including zero-width spaces and right-to-left override characters.
- Integration ready for defense-in-depth strategies within any Openclaw Skills automated agent.
SKILL.md
Loading
Related Openclaw Skills
OSOP: Open Standard Operating Procedures
OSOP provides a universal protocol for defining, validating, and self-optimizing structured workflows for AI agents.
OSOP Session Logger
A specialized logging utility that generates structured OSOP workflow definitions and execution records for AI agent sessions.
OSOP Workflow Reviewer
A specialized security audit tool designed to analyze AI agent workflow definitions and execution logs for risks, permission gaps, and destructive actions.
OSOP Report Generator
A specialized utility to transform complex workflow definitions and execution logs into beautiful, interactive HTML reports.
MyMBTI Personality Style Switcher
A dynamic communication skill that pivots AI agent dialogue styles across 16 MBTI personality frameworks based on user intent.
Tallow Candle Making Basics
A specialized AI agent skill for converting animal fat into high-quality, long-burning candles through precise chemical ratios and safety protocols.
Featured*
