[Naveen Chatlapalli (1aifanatic)]

Motivation & Context

Five of the eight code examples in agent_framework/_serialization.py fail when run:

Example	Error on main
SerializationProtocol: "simpler classes like UsageDetails"	AttributeError: 'dict' object has no attribute 'to_dict' (UsageDetails is a TypedDict)
SerializationMixin: "Framework tools with exclusion patterns"	ImportError: cannot import name 'BaseTool' from 'agent_framework._tools'
SerializationMixin: CustomAgent	TypeError: BaseAgent.__init__() got an unexpected keyword argument 'context_provider', and CustomAgent.from_dict(...) then fails with got multiple values for keyword argument 'name'
from_dict: OpenAI client injection	ValueError: Type mismatch: expected 'open_a_i_chat_client', got 'open_ai_chat_client'. The injected key is also client, but the constructor parameter is async_client
from_dict: AgentContext injection	AttributeError: type object 'AgentContext' has no attribute 'from_dict'

Description & Review Guide

• What are the major changes? Docstring examples only:
  • The protocol example uses Content, which does implement to_dict/from_dict.
  • The exclusion example uses FunctionTool, which has additional_properties in DEFAULT_EXCLUDE and func in INJECTABLE.
  • CustomAgent uses context_providers and sets its name and description defaults through kwargs.setdefault, so from_dict can pass them back in.
  • The OpenAI example uses the real type id open_a_i_chat_client and injects async_client.
  • The AgentContext example, which isn't serializable, becomes an example of the documented instance-specific injection ("name:get_weather" keys).
• What is the impact of these changes? Documentation only.
• What do you want reviewers to focus on? open_a_i_chat_client is what _get_type_identifier() produces for OpenAIChatClient. If that identifier isn't intended, it's worth a separate issue, but I left the code alone here.

I extracted and ran every example in the module: all 8 run on this branch. I also checked their outputs: the restored agent keeps its name, each tool receives the function meant for it, and the injected client is the one that was passed in.

Related Issue

N/A - trivial change: docstring examples are corrected to the current APIs, with no behavior change.

AI Assistance

• No material AI assistance was used.
• This is an AI-assisted contribution. I reviewed, understood, and verified all submitted content and accept responsibility for it.

AI assistance details: Claude Code helped find the failing examples by executing every docstring code block, rewrite them, re-run them, and draft this description.

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR links to an agreed issue with no competing open PR, or the Related Issue section documents a trivial-change or repository-automation exception.
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and title prefix in sync automatically.

🤖 Generated with Claude Code

[Copilot]

🟢 Approval recommended

No active review issues are reported for this documentation-only update to serialization examples.

No open issues 🎉

What changed in this PR

This documentation-only PR corrects serialization examples to match current Agent Framework APIs. Runtime behavior is unchanged.

Changes:

• Replace unsupported UsageDetails, BaseTool, and AgentContext examples with serializable types and instance-specific tool injection.
• Correct CustomAgent initialization and its documented serialized fields.
• Fix the OpenAI client type identifier and dependency injection parameter.

File	Description
python/​packages/​core/​agent_framework/​_serialization.py	Corrects five docstring examples covering serialization, exclusions, and dependency injection.

_{Review effort: Balanced}