Researchers have identified 26 third-party AI LLM routers that are secretly injecting malicious tool calls and stealing credentials. This discovery poses a significant risk to developers using AI coding agents, such as Claude Code, particularly those working on smart contracts or cryptocurrency wallets. The compromised routers could lead to the exposure of private keys and seed phrases, potentially endangering digital assets. The findings highlight the importance of vigilance among developers utilizing AI tools in the cryptocurrency space. Ensuring the security of development environments and verifying the integrity of AI tools are crucial steps to protect sensitive information from being compromised by these malicious actors.