Upgrade agent-framework to latest RC

[pamelafox]

Summary

This PR updates the Microsoft Agent Framework examples to the latest API shape and refreshes the related dependencies.

What changed

• Upgrade agent-framework to >=1.0.0rc5
• Refresh uv.lock for the updated Agent Framework dependency graph
• Migrate the Agent Framework examples from the older ChatAgent pattern to Agent
• Switch the Azure client setup in the Agent Framework examples to the OpenAIChatClient + async DefaultAzureCredential / get_bearer_token_provider pattern
• Register MCP tools on the agent (tools=[mcp_server]) instead of passing tools at run(...) time
• Update example response handling to print result.text
• Update default model IDs in the examples to gpt-4.1-mini
• Remove the Tavily examples

Tavily removal

The Tavily examples were deleted because the Tavily API key was not being handled in a secure way for these samples, and the Tavily examples were not referenced from the README or slide materials.

Files updated

• agents/agentframework_http.py
• agents/agentframework_learn.py
• pyproject.toml
• uv.lock

Files removed

• agents/agentframework_tavily.py
• agents/langchainv1_tavily.py

Notes

This branch also includes a merge from upstream/main. If GitHub shows additional changes relative to this repository's main, those are from bringing the branch up to date with upstream before applying the Agent Framework migration.

Validation

• uv sync

The Tavily examples were removed after confirming they were not referenced by the project README or slide-like documentation files.

[Copilot]

Pull request overview

This PR upgrades the project’s agent-framework dependency to the 1.0.0rc5 release candidate and updates the example agent scripts to align with the newer agent-framework API and dependency graph.

Changes:

• Bump agent-framework from 1.0.0b251016 to 1.0.0rc5 in pyproject.toml and refresh uv.lock.
• Update agentframework_* examples to use Agent + OpenAIChatClient (including async Azure credential/token provider flow).
• Adjust Tavily MCP configuration to pass the Tavily API key via URL query parameters.

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
uv.lock	Updates lockfile for agent-framework==1.0.0rc5, introducing new transitive deps/integrations.
pyproject.toml	Bumps project dependency requirement to agent-framework>=1.0.0rc5.
agents/langchainv1_tavily.py	Updates Tavily MCP connection URL construction (now includes API key in query string).
agents/agentframework_tavily.py	Migrates to new agent-framework API and updates Tavily MCP usage.
agents/agentframework_learn.py	Migrates to new agent-framework API and updates MCP tool usage.
agents/agentframework_http.py	Migrates to new agent-framework API and updates MCP tool usage / long-running behavior.

Comments suppressed due to low confidence (1)

agents/agentframework_tavily.py:72

• async_credential.close() is only reached on the success path; if an exception occurs during MCP/tool setup or agent.run(), the credential won't be closed. Wrap the agent execution in a try/finally (or use an async context manager, if available) to ensure the Azure credential is always closed.

    if async_credential:
        await async_credential.close()

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 7 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

AZURE_OPENAI_ENDPOINT in .env-sample includes a trailing slash, so base_url=f"{os.environ['AZURE_OPENAI_ENDPOINT']}/openai/v1/" will end up with //openai/v1/. Consider stripping any trailing / from the endpoint before appending /openai/v1/.

Suggested change

	client = OpenAIChatClient(
	base_url=f"{os.environ['AZURE_OPENAI_ENDPOINT']}/openai/v1/",
	azure_endpoint = os.environ["AZURE_OPENAI_ENDPOINT"].rstrip("/")
	client = OpenAIChatClient(
	base_url=f"{azure_endpoint}/openai/v1/",

[Copilot]

Depending on the agent-framework meta-package now pulls in agent-framework-core[all] (see uv.lock), which brings a large set of optional integrations and heavy transitive deps (e.g., pythonnet, durabletask, etc.). If these examples only need the core OpenAI client + MCP tool support, consider depending on agent-framework-core (or a smaller extra set) instead to keep installs lighter and reduce platform-specific dependency risk.

Suggested change

	"agent-framework>=1.0.0rc5",
	"agent-framework-core>=1.0.0rc5",

[Copilot]

AZURE_OPENAI_ENDPOINT in .env-sample ends with a trailing slash, so building base_url as f"{os.environ['AZURE_OPENAI_ENDPOINT']}/openai/v1/" produces a double-slash (...azure.com//openai/v1/). This can break request routing in some HTTP stacks. Consider normalizing the endpoint first (e.g., rstrip('/')) before appending the path.

Suggested change

	client = OpenAIChatClient(
	base_url=f"{os.environ['AZURE_OPENAI_ENDPOINT']}/openai/v1/",
	azure_endpoint = os.environ["AZURE_OPENAI_ENDPOINT"].rstrip("/")
	client = OpenAIChatClient(
	base_url=f"{azure_endpoint}/openai/v1/",

[pamelafox]

I'm okay with those comments.