agentproof

What This Skill Does

The AgentProof skill allows OpenClaw users to interface with the AgentProof Oracle, a comprehensive trust-scoring engine designed specifically for ERC-8004 agents. By leveraging an 8-factor proprietary algorithm, the tool provides real-time verification of agent credibility across Ethereum, Avalanche, Base, and Linea networks. It offers granular insights into an agent's historical performance, deployment patterns, and risk profile, ensuring users interact only with reputable entities. This helps prevent financial loss from fraudulent agents and avoids the execution of malicious or unreliable code.

Installation

To integrate this safety tool into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/builderbenv1/agentproof Ensure you have the latest version of the ClawHub CLI installed to manage repository dependencies and verify the digital signature of the package.

Use Cases

• Pre-Payment Verification: Use this skill before processing x402 payments to verify if an agent has a history of delivering promised tasks.
• Dependency Security: Scrutinize unknown agents before importing or installing skills that originated from their URI endpoints.
• Agent Comparison: When evaluating multiple agent services for high-stakes tasks like financial management or data analysis, compare their leaderboard rankings and individual trust scores.
• Risk Mitigation: Automatically check for active risk flags like 'SERIAL_DEPLOYER' or 'NEW_IDENTITY' before delegating sensitive system access to an automated agent.

Example Prompts

1. "AgentProof, check the trust score for agent 0x71C... and tell me if they are safe to hire for my web scraping project."
2. "What is the current leaderboard for top-rated agents on the Base chain, and which ones have the lowest risk flags?"
3. "Run a risk assessment on the deployer of this agent. Do they have a history of abandoning their previous agents?"

Tips & Limitations

Always check for the 'NEW_IDENTITY' flag; agents under 7 days old lack sufficient data for a highly accurate score. Remember that while AgentProof uses a robust 8-factor model—including Rating Score, Consistency, and URI Stability—it is a decision-support tool. A high 'Gold' or 'Platinum' tier indicates strong past performance but cannot guarantee future behavior. Periodically check network stats via the skill to ensure the oracle is updated with the latest chain indexing data.