Add user-defined policies

[bhelx]

This is an experiment and meant to be discussed in the broader context of mcpx and the topic of agent control. The code is fine but is meant to be thrown away for something broader. Please ignore for now.

[chrisdickinson]

What do you think about folding this into a single intercept "middleware" function?

mcpx.intercept("fetch", async (request, context, next) => {
  const { hostname } = new URL(request.params.arguments.url)
  const counts = context.get("fetch.domain-counts") || {}
  if (!context.has("fetch.domain-counts")) {
    context.set("fetch.domain-counts", counts)
  }

  counts[hostname] ??= 0
  // we can only call each domain 3 times for some reason!
  // our lawyers are making us!
  if (counts[hostname] >= 3) {
    console.log(`failed`)
    return {
      isError: true,
      text: `The domain ${hostname} was called too many times: ${currentCount}`
    }
  }

  const result = await next(request, context)

  ++counts[hostname]
  console.log({hostname, counts, count})

  return result
})

I.e., instead of an allow/deny response, give the library the ability to control handling the incoming tool call request or delegating it to the next layer down. This makes it possible to stub tool invocations during tests, call a tool multiple times, return its own response in special circumstances, etc. (You could implement a policy API on top of this.)