Master the practical building blocks of agentic systems in TypeScript. Covering Factors 1, 3, 4, 8, and 9, this course teaches structured model outputs with the official OpenAI TypeScript SDK, typed tool schemas, explicit context management, and controlled loops that execute tools and compact errors back into context.
Overview
Syllabus
Structured Outputs in TypeScript
- Make a Basic Responses API Call
- Build a Product Categorizer with a System Prompt
- Parsing Structured City Data
- Extracting and Printing To-Do Tasks
- Handle Invalid JSON Safely
Native Tool Calling
- Replace Prompted JSON with Native Tool Calls
- Require Tool Use with tool_choice
- Parse the Final Answer Tool Call
- Expanding a Tool Schema with Confidence
- Route Multiple Tool Calls
The Tool Execution Loop
- Defining Tool Schemas for Math Functions
- Execute the Model's Tool Calls
- Managing Context During Tool Execution
- Complete the Tool Execution Loop
- Supporting Parallel Tool Calls
Building Agentic Loops
- Build the Agent Loop Control Skeleton
- Inspecting Model Tool Calls in the Agent Loop
- Complete the Agent Execution Cycle
- Add Post-Loop Summary Checks
- Extend the Agentic Loop for a More Complex Calculation
