trustless-work-dev

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill configures MCP servers that the agent will call at runtime—specifically "https://docs.trustlesswork.com/trustless-work/~gitbook/mcp" and "https://mcp.trustlesswork.com/mcp"—which stream documentation and expose tools to trigger live operations, so these URLs are runtime dependencies that can directly control agent behavior.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly provides REST endpoints, SDK hooks, and MCP integration for escrow operations that move funds: deploy, fund-escrow, approve, release, dispute/resolve, and helper/send-transaction to broadcast signed transactions to Stellar. It includes API base URLs and examples of the full fund → sign → submit flow and React hooks like useFundEscrow, useReleaseFunds, and useSendTransaction, so it is specifically designed to execute financial transactions on-chain.