AI systems 0/10

Lesson 7 of 10

Agents and tool use

Run a model in a loop that calls typed tools, trace one call end to end, connect tools over MCP, and add a step budget and approval for risky actions.

1,831 words · 9 min read Ultra Premium Useful background: How an LLM call works, API design

In this lesson

  1. The agent loop
  2. One tool call, end to end
  3. MCP connects tools
  4. Separate proposal and effect
  5. Execution states
  6. Drive the bounded loop
  7. Implement the executor
  8. What stays true
  9. Cost and tradeoffs
  10. Common mistakes
  11. Related topics and practice

Ultra Premium

Unlock this lesson

Ultra Premium opens the 10 premium lessons in AI systems, along with the rest of WhiteBox's premium lessons and problems.

Not ready yet? Read the free lesson: Backpressure