🛡️ agent-safe-pipeline: AI Agent Execution Authority Gateway

As AI agents gain the ability to execute real-world actions, security and permission control become critical pain points. agent-safe-pipeline introduces an execution-authority gateway designed specifically to intercept and audit consequential HTTP requests made by agents. This is not just a standard API router; it acts more like a checkpoint with strict audit trails. When an agent attempts a sensitive operation, the gateway intercepts the request, verifies authorization, or holds it for human approval. Every action leaves a chained evidence trail, and grants are strictly single-use. The hard part of agentic AI is not getting models to make requests, but safely constraining them. This project offers an interesting infrastructure-level approach to enterprise AI governance and agent safety.