1
Curious builder0 XP earned · 300 to level 2
0 daysFinish a lesson to begin
Badge collection0 of 6 unlocked
27 small wins to finish your path
What we left out, and what is next
This course taught the SDK's core; several teachable areas were left out, listed here with a doc link each so you know where to read next.
Last updated: 28 Sep, 2026 · openai-agents 0.22.3
You built an agent, gave it tools and structured output, handed off to a specialist, guarded its input and output, gave it memory, streamed and traced a run, capped it, and assembled a shop support desk. The SDK covers more than one course can. The table names what was left out and where the docs cover it.
Teachable areas this course did not cover
| Area | What it adds | Docs |
|---|---|---|
| Hosted tools (WebSearchTool, FileSearchTool, CodeInterpreterTool) | Tools that run on OpenAI's servers, no function of yours needed | /tools/ |
| ComputerTool | Drives a browser or computer through a local runtime you supply | /tools/ |
| MCP servers | Connect external tool servers over the Model Context Protocol | /mcp/ |
| Voice | Turn agents into a speech-in, speech-out pipeline | /voice/quickstart/ |
| Realtime | Low-latency streaming audio agents | /realtime/quickstart/ |
| Lifecycle hooks | Callbacks on run, agent, tool, and handoff events | /ref/lifecycle/ |
| Visualization | Draw the agent and handoff graph as a diagram | /visualization/ |
| Model providers | Run non-OpenAI models through the same SDK | /models/ |
| Human-in-the-loop / tool approval | Pause a run so a person approves or rejects a tool call before it runs | docs home |
| Fault tolerance and retries | Retrying failed model or tool calls and recovering a run after an error | /running_agents/ |
Where to read next
- Start with the docs home at openai.github.io/openai-agents-python for the full section list.
- For a bigger build, add hosted tools or an MCP server to the desk you finished in A full run, and when it refuses.
- For production, read the tracing and lifecycle pages to watch and shape a live run.
Watch out. Hosted tools and voice run against real OpenAI services, so they need
OPENAI_API_KEY and are billed, unlike the stand-in that carried this course.Related
- Previous: Swapping in a real model
- Reference: OpenAI Agents SDK docs
Try it yourself
- Open the docs page for one left-out topic and read its first example.
- Pick the next thing to add to your desk agent from the list above.
Little by little, you're building something great.