sardis-payment

What This Skill Does

The sardis-payment skill integrates the Sardis Payment OS directly into your OpenClaw AI agent, transforming it into a secure, policy-governed financial entity. It bridges the gap between autonomous decision-making and real-world capital by providing non-custodial MPC wallets. The skill enables agents to execute payments across major chains including Base, Polygon, Ethereum, Arbitrum, and Optimism using stablecoins like USDC and USDT. Unlike traditional agentic payment flows, sardis-payment mandates a 'policy-first' approach, ensuring every transaction is audited, validated against natural language spending rules, and recorded for compliance. It supports card management, real-time balance monitoring, and automated transaction logging.

Installation

To integrate the skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/efedurmaz16/sardis-openclaw Ensure you have configured your environment variables by setting SARDIS_API_KEY and SARDIS_WALLET_ID to allow the agent authorized access to your wallet infrastructure. Refer to the Sardis developer dashboard to generate your keys securely.

Use Cases

• Automated SaaS Procurement: Have your agent subscribe to and manage API credits (e.g., OpenAI, Anthropic, Midjourney) by monitoring usage and paying invoices automatically based on strict budget caps.
• Decentralized Procurement: Automatically purchase compute power or data services on-chain without human intervention, ensuring the vendor address is verified and the transaction is within defined spend thresholds.
• Budget Compliance Enforcement: Delegate budget-constrained tasks to agents that must verify if an expense is permitted by company policy before committing funds.

Example Prompts

1. "Check if I have enough balance on my Base wallet to cover a $50 payment to the OpenAI API and let me know if it violates my policy."
2. "Execute a payment of 10 USDC to the following address on Polygon: 0x123...456, but only if the policy allows for general software subscriptions."
3. "Show me the last 5 transactions on my wallet and calculate the total spent on infrastructure this week."

Tips & Limitations

• Always use dry-run: Before executing any transfer, use the policy check endpoint to avoid unexpected rejections or failed transaction costs.
• Failure Protocol: The system is designed to 'fail closed.' If your policy settings are too restrictive, the agent will refuse the payment. Ensure your natural language policies are clearly defined.
• Security: Never hardcode credentials in your agent scripts. Use environment variables. The skill does not support private key exports for security reasons; all signing is handled by the MPC backend.