ClawShell provides a robust security layer for Openclaw Skills by intercepting shell commands and requiring manual human approval for high-risk operations.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawshell
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 clawshell 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 ClawShell?
ClawShell serves as a professional-grade security middleware designed to protect your environment while using Openclaw Skills. It functions as a human-in-the-loop gatekeeper that analyzes every shell command an AI agent attempts to execute, categorizing them into risk levels ranging from Low to Critical.
By prioritizing safety, this skill ensures that potentially destructive actions like accidental file deletions or unauthorized network access are impossible without explicit verification. It is a vital component for developers who want the efficiency of AI agents with the peace of mind of human oversight.
ClawShell Use Cases
- Guarding against accidental execution of destructive commands like rm -rf in production or development environments.
- Controlling external network access by intercepting curl or wget requests to untrusted URLs.
- Implementing a secure review process for git push operations and automated dependency installations.
- Maintaining a comprehensive, tamper-evident audit log of all agent-driven terminal activity.
How ClawShell Works
- The AI agent invokes the clawshell_bash tool instead of the standard system bash shell.
- ClawShell performs an immediate risk analysis by comparing the command against built-in heuristics and custom blocklists.
- If the command is deemed high-risk, a push notification is sent to the user via Pushover or Telegram.
- The execution lifecycle enters a waiting state, blocking further action until the user provides a 'Permit' or 'Deny' response.
- Low-risk commands are executed immediately, while critical threats are automatically blocked without user intervention to prevent immediate damage.
ClawShell Setup
To integrate this security layer into your Openclaw Skills setup, follow these steps:
- Navigate to the skill directory and install the required Node.js dependencies:
cd /app/workspace/skills/clawshell
npm install
- Configure your notification provider in the
.envfile to receive approval requests:
CLAWSHELL_PUSHOVER_USER=your-user-key
CLAWSHELL_PUSHOVER_TOKEN=your-app-token
- Update your agent's TOOLS.md to instruct the agent to use
clawshell_bashfor all shell-related tasks, ensuring a secure execution environment.
ClawShell Data Schema & Taxonomy
ClawShell organizes its audit data using a structured JSONL format to ensure high performance and easy parsing for security audits. All activities are recorded in the directory specified by the CLAWSHELL_LOG_DIR environment variable.
| Log Property | Description |
|---|---|
timestamp |
ISO 8601 formatted string of the event time |
command |
The raw shell command intercepted for analysis |
risk_level |
Classification based on impact (Low, Medium, High, Critical) |
decision |
The final outcome of the request (Allowed, Blocked, Approved, Rejected) |
output |
Captured stdout and stderr from the command if executed |
ClawShell Advanced Features
- Multi-channel notifications with native support for both Pushover and Telegram bot integration.
- Flexible security policies via config.yaml, allowing for sophisticated glob and regex-based command filtering.
- Configurable execution timeouts to prevent the agent from hanging indefinitely while waiting for human input.
- Human-in-the-loop status tools to query pending approvals and recent decision history directly from the agent.
SKILL.md
Loading
Related Openclaw Skills
Quests
A standardized framework for AI agents to track and guide humans through complex, multi-day real-world processes.
API Token Speed Benchmark
A professional benchmarking utility to measure and compare token generation speed, latency, and throughput across multiple LLM API providers.
Lyric Translator: Indonesian to English
A specialized AI agent skill designed to translate Indonesian song lyrics into authentic, singable, and human-sounding English for international music production.
Fitbit Fitness Insights
A powerful AI integration for Fitbit that provides deep insights into your physical activity, sleep patterns, and heart rate data.
Verfi
A compliance-focused tool for capturing and verifying tamper-proof TCPA consent sessions during the lead generation process.
Renzo
Query live data from the Renzo liquid restaking protocol, including ezETH metrics, vault performance, and user portfolio balances.
Featured*
