AI Agents and Tool Use

Learn how AI agents use tools: ReAct and planning, tool schemas, tool design, safety, evaluation and multi-agent patterns, with working Python examples.

Start course →

Syllabus

Agent Foundations

  1. The Spectrum of Autonomy
  2. ReAct: Reason, Act, Observe
  3. Planning: Reactive or Plan-and-Execute
  4. Reflection and Self-Checking

Tool Use Essentials

  1. Tool Schemas
  2. A Tool Registry in Code
  3. tool_choice and Parallel Calls
  4. Structured Output via Tools

Designing Good Tools

  1. Granularity and Naming
  2. Descriptions and Parameter Docs
  3. Side Effects and Idempotency
  4. Errors That Guide the Model

Kinds of Tools

  1. Retrieval and Search Tools
  2. Code Execution Tools
  3. API, Database and Computer-Use Tools

Safety for Tool-Using Agents

  1. Risk Tiers and Approvals
  2. Least Privilege and Sandboxing
  3. Prompt Injection via Tool Results

Reliability and Evaluation

  1. Evaluating Trajectories
  2. Building an Eval Set
  3. Human-in-the-Loop Design

Multi-Agent Systems and MCP

  1. When Multiple Agents Help
  2. MCP from the Tool Author's View

Build and Revise

  1. Case Study: An Expense Assistant
  2. Revision: Cheat Sheet and Self-Check