Practical security guidelines for developers using AI coding assistants — token scopes, secrets management, sandboxing, prompt injection, and reviewing generated code.
