| layout | default |
|---|---|
| title | Chapter 4: Agents and Assistants |
| nav_order | 4 |
| parent | OpenAI Python SDK Tutorial |
Welcome to Chapter 4: Agents and Assistants. In this part of OpenAI Python SDK Tutorial: Production API Patterns, you will build an intuitive mental model first, then move into concrete implementation details and practical production tradeoffs.
This chapter focuses on transition strategy: operate existing assistants safely while moving toward current agent-platform patterns.
- Assistants API is still usable in many systems.
- OpenAI platform docs indicate a target sunset timeline around August 26, 2026.
- New projects should evaluate Responses API + Agents patterns first.
from openai import OpenAI
client = OpenAI()
assistant = client.beta.assistants.create(
model="gpt-5.2",
name="Ops Assistant",
instructions="Help with reliability planning and incident response."
)
thread = client.beta.threads.create()
client.beta.threads.messages.create(
thread_id=thread.id,
role="user",
content="Draft a rollback checklist for a risky deployment."
)- catalog Assistants API usage and tool dependencies
- extract shared prompt/tool contracts
- rebuild core flows on Responses/Agents primitives
- run side-by-side output comparisons
- cut over service by service
- avoid broad rewrites in one release
- pin SDK versions per service
- keep rollback path to known-good behavior
- monitor quality regressions with fixed eval sets
You can now manage assistant-era systems while executing a controlled migration plan.
Next: Chapter 5: Batch Processing
The public_members function in scripts/detect-breaking-changes.py handles a key part of this chapter's functionality:
def public_members(obj: griffe.Object | griffe.Alias) -> dict[str, griffe.Object | griffe.Alias]:
if isinstance(obj, griffe.Alias):
# ignore imports for now, they're technically part of the public API
# but we don't have good preventative measures in place to prevent
# changing them
return {}
return {name: value for name, value in obj.all_members.items() if not name.startswith("_")}
def find_breaking_changes(
new_obj: griffe.Object | griffe.Alias,
old_obj: griffe.Object | griffe.Alias,
*,
path: list[str],
) -> Iterator[Text | str]:
new_members = public_members(new_obj)
old_members = public_members(old_obj)
for name, old_member in old_members.items():
if isinstance(old_member, griffe.Alias) and len(path) > 2:
# ignore imports in `/types/` for now, they're technically part of the public API
# but we don't have good preventative measures in place to prevent changing them
continue
new_member = new_members.get(name)
if new_member is None:
cls_name = old_member.__class__.__name__
yield Text(f"({cls_name})", style=Style(color="rgb(119, 119, 119)"))
yield from [" " for _ in range(10 - len(cls_name))]This function is important because it defines how OpenAI Python SDK Tutorial: Production API Patterns implements the patterns covered in this chapter.
The find_breaking_changes function in scripts/detect-breaking-changes.py handles a key part of this chapter's functionality:
def find_breaking_changes(
new_obj: griffe.Object | griffe.Alias,
old_obj: griffe.Object | griffe.Alias,
*,
path: list[str],
) -> Iterator[Text | str]:
new_members = public_members(new_obj)
old_members = public_members(old_obj)
for name, old_member in old_members.items():
if isinstance(old_member, griffe.Alias) and len(path) > 2:
# ignore imports in `/types/` for now, they're technically part of the public API
# but we don't have good preventative measures in place to prevent changing them
continue
new_member = new_members.get(name)
if new_member is None:
cls_name = old_member.__class__.__name__
yield Text(f"({cls_name})", style=Style(color="rgb(119, 119, 119)"))
yield from [" " for _ in range(10 - len(cls_name))]
yield f" {'.'.join(path)}.{name}"
yield "\n"
continue
yield from find_breaking_changes(new_member, old_member, path=[*path, name])
def main() -> None:
try:
against_ref = sys.argv[1]This function is important because it defines how OpenAI Python SDK Tutorial: Production API Patterns implements the patterns covered in this chapter.
The main function in scripts/detect-breaking-changes.py handles a key part of this chapter's functionality:
def main() -> None:
try:
against_ref = sys.argv[1]
except IndexError as err:
raise RuntimeError("You must specify a base ref to run breaking change detection against") from err
package = griffe.load(
"openai",
search_paths=[Path(__file__).parent.parent.joinpath("src")],
)
old_package = griffe.load_git(
"openai",
ref=against_ref,
search_paths=["src"],
)
assert isinstance(package, griffe.Module)
assert isinstance(old_package, griffe.Module)
output = list(find_breaking_changes(package, old_package, path=["openai"]))
if output:
rich.print(Text("Breaking changes detected!", style=Style(color="rgb(165, 79, 87)")))
rich.print()
for text in output:
rich.print(text, end="")
sys.exit(1)
main()This function is important because it defines how OpenAI Python SDK Tutorial: Production API Patterns implements the patterns covered in this chapter.
flowchart TD
A[public_members]
B[find_breaking_changes]
C[main]
A --> B
B --> C