A secure sudo-protocol for AI agents that requires manual human approval before executing destructive, financial, or infrastructure-level actions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install letsping
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 letsping 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 letsping?
letsping is a critical safety layer designed for autonomous agents, implementing a human-in-the-loop approval workflow for sensitive operations. By acting as a secure escalation point, it ensures that your agent does not perform high-risk tasks—such as modifying production infrastructure or executing financial transactions—without explicit authorization. This skill is a vital addition to any professional implementation of Openclaw Skills, offering a robust mechanism to maintain human oversight over autonomous logic.
Built on the sudo protocol, letsping forces the agent to call a specialized tool whenever a risk threshold is met. This tool blocks execution and notifies the user via a Progressive Web App (PWA), where the payload can be reviewed, edited, or rejected. Only after a human provides the green light does the agent receive the authorized payload to proceed, significantly reducing the surface area for autonomous errors or prompt injection exploits.
letsping Use Cases
- Preventing unauthorized financial transactions or bookings that exceed a defined budget.
- Protecting cloud infrastructure by requiring approval for DNS changes or deployment triggers.
- Ensuring data integrity by intercepting destructive commands like file deletions or database overwrites.
- Moderating social communications to prevent accidental data leaks or unauthorized public posts.
How letsping Works
- The AI agent identifies an intended action that matches a high-risk category defined in its system prompt.
- The agent calls the letsping_ask tool, passing the original tool name, the intended JSON arguments, and a justification for the risk.
- The skill pauses execution and transmits the request to the user via the LetsPing secure gateway.
- The human operator reviews the request in the LetsPing PWA, choosing to approve the action as-is, patch the arguments for safety, or reject the request entirely.
- The tool returns the user-validated payload to the agent, which then executes the authorized version of the command.
letsping Setup
To integrate this safety layer into your Openclaw Skills setup, follow these installation steps:
npm install @letsping/openclaw-skill
Alternatively, you can clone the repository directly into your local workspace:
git clone https://github.com/CordiaLabs/openclaw-skill ~/.openclaw/workspace/skills/letsping
cd ~/.openclaw/workspace/skills/letsping
npm install
Finally, configure your environment by setting the LETSPING_API_KEY. You can obtain this key by pairing your agent at the official letsping website.
letsping Data Schema & Taxonomy
The letsping skill handles structured data to ensure transparency during the human-in-the-loop process. Below is the schema used for escalation requests:
| Property | Type | Description |
|---|---|---|
tool_name |
String | The exact name of the high-risk tool the agent intends to use. |
args_json |
String | A stringified JSON object containing the original arguments for the tool. |
risk_reason |
String | A clear explanation of the potential risk (e.g., 'Destructive file deletion'). |
status |
String | The outcome of the human review (approved, rejected, or patched). |
executed_payload |
JSON | The final, authorized payload returned to the agent for execution. |
letsping Advanced Features
- Real-time payload patching, allowing humans to modify agent arguments on the fly before execution.
- Multi-category risk detection covering financial, destructive, social, and infrastructure domains.
- Compatibility with frontier LLMs to minimize the risk of prompt bypass and ensure strict protocol adherence.
- Secure integration with the LetsPing PWA for cross-device approval notifications.
SKILL.md
Loading
Related Openclaw Skills
Meeting Minutes Assistant
An automated workflow for generating comprehensive meeting minutes and actionable tasks from raw meeting data.
Flight Tracker
A comprehensive flight tracking tool for monitoring real-time aviation data including delays, gates, and live positions.
Flight Tracker
A real-time flight monitoring tool that provides detailed status updates, gate information, and live position data for worldwide aviation.
Email Reader
A comprehensive skill enabling AI agents to search, filter, and retrieve email content from Gmail and IMAP accounts via command-line interfaces.
Baidu Map Integration
Integrate comprehensive Baidu Map Web Service APIs for location searching, geocoding, and routing.
Cross-Device Sync Skill
A synchronization tool that uses GitHub as a central repository to keep OpenClaw memory files consistent across all your devices.
Featured*
