With any agent sandbox or client that you use for interacting with LLMs, the same question will always arise: how can I securely, and in an observable fashion, connect to endpoints (LLMs, MCP
Think about two scenarios that are pretty common. 1) You hit a rate limit or run out of tokens, so you have to "downgrade" to a small/less powerful Model. 2)
"treat 'em like cattle, not pets".
This was, and continues to be, how many look at Kubernetes Pods and microservice-based architecture. It makes a lot of sense for objects like
An Agent makes a call to an LLM. The LLM decides which MCP server tool should be used for a task. The Agent then makes a call to said tool. This can happen
Your Agent has a "mind of its own" (well, it was programmed to act a particular way). For example, Claude Code is known to downgrade your Model for particular tasks to
