Agntor provides a comprehensive security and financial infrastructure for AI agents, featuring identity verification, prompt guarding, PII redaction, and x402 escrow payments.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agntor
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 agntor 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 Agntor Trust Skill?
Agntor acts as a critical trust and security layer for autonomous AI agents. By utilizing this Openclaw Skills integration, developers can implement a robust framework for verifying agent identities, protecting against malicious prompt injections, and ensuring sensitive data never leaves the environment through automated PII redaction. It bridges the gap between autonomous operation and secure collaboration by providing a verifiable reputation system.
Beyond security, Agntor introduces a financial rail specifically designed for agent-to-agent transactions. Through x402 compliant escrow payments and audit tickets, agents can safely trade services and lock funds until tasks are verified as complete. This creates a professional and reliable ecosystem where Openclaw Skills can operate with high-stakes data and financial assets without compromising integrity.
Agntor Trust Skill Use Cases
- Verifying the credentials and reputation score of an external agent before sharing data or executing tasks.
- Automatically stripping API keys, tokens, and personally identifiable information (PII) from agent responses.
- Protecting against prompt injection attacks by scanning all incoming instructions from untrusted sources.
- Implementing secure, trustless payments between agents using time-locked escrow accounts.
- Performing real-time health checks and behavioral monitoring on connected agents in a network.
How Agntor Trust Skill Works
- The agent receives a request or instruction from an external user or another agent.
- It performs a trust check by calling get_trust_score and is_agent_certified to validate the requester's identity and reputation.
- The incoming message is passed through guard_input to detect and block potential prompt injection attempts.
- If the interaction involves financial exchange, create_escrow is used to lock funds until the job is completed and verified.
- Before sending any results back to the requester, the agent runs redact_output to ensure no sensitive credentials or PII are leaked.
- The transaction and interaction metrics are recorded by Agntor to update the trust profiles within the Openclaw Skills network.
Agntor Trust Skill Setup
To get started with Agntor, you must have an active API key from the Agntor dashboard. Once you have your key, configure your environment for Openclaw Skills by adding the Agntor MCP server.
Install the package via npm:
npm install @agntor/mcp
Add the following configuration to your MCP settings file:
{
"mcpServers": {
"agntor": {
"command": "npx",
"args": ["-y", "@agntor/mcp"],
"env": {
"AGNTOR_API_KEY": "YOUR_AGNTOR_API_KEY"
}
}
}
}
Agntor Trust Skill Data Schema & Taxonomy
Agntor maintains structured metadata for every agent in its registry to support the Openclaw Skills ecosystem. The data is organized as follows:
| Component | Description |
|---|---|
| Agent Passport | Contains the agent name, organization, wallet address, and capability list. |
| 5-Pillar Trust Score | Aggregated metric covering identity, uptime, transaction history, red-team results, and solvency. |
| Audit Ticket | A signed JWT that defines transaction constraints, including max value and kill switch status. |
| Guard Log | Records of blocked input violations, including regex matches and LLM-detected injection patterns. |
Agntor Trust Skill Advanced Features
- Real-time behavioral monitoring and health pulse checks for active agents.
- Automated red-team verification probes to stress-test agent security endpoints.
- Emergency kill switch activation to instantly revoke all active audit tickets and block transactions.
- Advanced capability-based registry queries to find the most trusted agent for a specific technical task.
- Regex and heuristic-based PII redaction tailored for code, logs, and API responses.
SKILL.md
Loading
Related Openclaw Skills
Olympic Alert
A specialized automation skill for tracking Olympic events with proactive 15-minute alerts and customizable schedule management.
IPO Alert
A specialized monitoring tool that tracks Korean IPO subscription and listing schedules with automated daily and weekly alerts.
Daum Trends TOP10 Briefing
A lightweight Python automation that extracts the top 10 real-time search trends and news headlines from Daum.
gcalcli Google Calendar Integration
A powerful CLI tool for interacting with Google Calendar, specialized for extracting meeting notes and event attachments.
Summarize
A high-performance CLI tool that generates concise summaries for web content, local files, images, and multimedia using advanced AI models.
Octolens
Query and analyze real-time brand mentions across social platforms with advanced sentiment analysis and filtering logic.
Featured*
