Add Toolset Authentication per issue #1227 - 4

[zyantw]

Added documentation for the new Toolset-level authentication feature and improved the visual rendering of our authentication flow diagrams.

• Staged: https://deploy-preview-2138--adk-docs-preview.netlify.app/tools-custom/authentication/
• Agent's PRs: Update ADK doc according to issue #1227 - 4 - Toolset Authentication #1233 and Update ADK doc according to issue #1227 - 4 - Toolset Authentication #1234
• Original issue: Found docs updates needed from ADK python release v1.23.0 to v1.24.0 #1227

---

Technical Verification Against google/adk-python

Verification of Import Statements

• Snippet Under Review:

  from google.adk.auth import AuthConfig
  from google.adk.tools.base_toolset import BaseToolset

• Status: PASS
• Citations & Verification:
  • auth/init.py:L25: AuthConfig is exposed directly by the google.adk.auth top-level namespace.
  • base_toolset.py:L38: BaseToolset is located at google.adk.tools.base_toolset. Importing from this module path resolves without error.

---

Method Signature and Subclassing

• Snippet Under Review:

  class MyAuthenticatedToolset(BaseToolset):
  
    def get_auth_config(self) -> AuthConfig | None:
      return AuthConfig(...)

• Status: PASS
• Citations & Verification:
  • base_toolset.py:L225-241: BaseToolset.get_auth_config defines def get_auth_config(self) -> Optional[AuthConfig]: return None.
  • The return type annotation AuthConfig | None is valid in Python 3.10+ and conforms with Optional[AuthConfig].
  • test_toolset_auth.py:L43-57: Unit tests demonstrate custom toolsets overriding get_auth_config() using this exact pattern.

---

Pre-Execution Authentication Resolution

• Draft Statement: "Under the hood, the BaseLlmFlow automatically checks your BaseToolset for authentication requirements before it even lists or executes any tools. It does this by checking the toolset's get_auth_config() method."
• Status: PASS
• Citations & Verification:
  • base_llm_flow.py:L1158-1167: In BaseLlmFlow._preprocess_async, self._resolve_toolset_auth(invocation_context, agent) executes before _process_agent_tools, guaranteeing credentials are resolved before get_tools() is called.
  • base_llm_flow.py:L171-177: _resolve_toolset_auth loops over agent.tools, inspects instances of BaseToolset, and queries tool_union.get_auth_config().

---

Pausing Execution and Emitting adk_request_credential

• Draft Statement: "1. Pause execution: It safely halts the current flow. 2. Request credentials: It issues an adk_request_credential event to the client..."
• Status: PASS
• Citations & Verification:
  • functions.py:L64: REQUEST_EUC_FUNCTION_CALL_NAME is set to 'adk_request_credential'.
  • functions.py:L341-351: build_auth_request_event builds the function call part using name=REQUEST_EUC_FUNCTION_CALL_NAME.
  • base_llm_flow.py:L220-227: Emits the event and sets invocation_context.end_invocation = True, safely interrupting the flow.

[netlify]

✅ Deploy Preview for adk-docs-preview ready!

Name	Link
🔨 Latest commit	4fd10b4
🔍 Latest deploy log	https://app.netlify.com/projects/adk-docs-preview/deploys/6a837583f2c93800084ba251
😎 Deploy Preview	https://deploy-preview-2138--adk-docs-preview.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[zyantw]

I tried my best to make the flow as close as possible to the original in the page. Do you happen to have the mermaid code for that one? I can't find it. @joefernandez