ClawSafe Pay – Agent Skill

You can route agent payment intents through ClawSafe Pay to enable Telegram-based human approval. The workflow automatically drafts transactions, performs an AI safety review, and waits for human authorization before signing and broadcasting.

A human-in-the-loop workflow for cross-chain payments ensures agents draft transactions but require human approval before execution. This skill uses policy checks, AI safety reviews, and Telegram authorization to maintain an auditable, secure event trail.

To initiate Sepolia testnet payments, your OpenClaw agent calls the ClawSafe Pay publisher service API with a payment intent. The service handles drafting, safety review, Telegram approval, signing, and broadcasting automatically.

ClawSafe Pay currently provides full support for Sepolia testnet payments only. Other cross-chain destinations are placeholders for future activation and are not yet available for production-ready transfers.

Automated agent payment workflows are secured using strict API key authentication and HMAC inter-service signing. This ensures only authorized agents submit payment intents to the publisher service while maintaining a verifiable event trail.

Yes, you need python-dotenv installed to manage environment configuration for the ClawSafe Pay agent skill. It allows you to securely load API keys and endpoint URLs required for publisher service authentication.