Add Google ADK agent example connecting to ToolHive vMCP
Nobody has claimed this yet.
- Dominant language
- Go
- Stars
- 2.2k
- Forks
- 300
- Avg merge
- 1d 15h
- Merged PRs (30d)
- 184
Description
User Story
As a ToolHive user attempting to build my own agent service with Google ADK, I have no
reference implementation showing how to securely connect a Google ADK agent to MCP
servers managed by ToolHive. Without this, I have to figure out the integration from
scratch.
Description
Create a self-contained Google ADK agent example in the toolhive repository that
demonstrates connecting a Google ADK agent to a ToolHive-managed vMCP endpoint with
Okta OIDC auth. Google ADK is gaining traction for building production agent services;
this example shows how ToolHive fits into that ecosystem.
Requirements
- Create
examples/agents/google-adk/directory with a runnable Python example - Demonstrate how to initiate the SSO flow from the agent service to authenticate to a vMCP
- Show the agent invoking at least one MCP tool through the vMCP
- Include a
README.mdwith prerequisites, setup steps, and how to run - Keep dependencies minimal and pinned
Acceptance Criteria
- Example is checked into
examples/agents/google-adk/in the toolhive repo - A user can follow the README and run the example against their own vMCP
- The example demonstrates successful MCP tool invocation through the vMCP
- The SSO flow is initiated from within the agent service
- README explains the ADK agent structure and how MCP is wired in
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Create the self-contained runnable example under examples/agents/google-adk/ and include its README.md. Start by defining the Python agent, its ToolHive-managed vMCP connection, and the Okta OIDC SSO flow; done means a user can follow the prerequisites and setup steps, run the agent against their own vMCP, and invoke an MCP tool successfully.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, security
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 52/100