A skill for the Gemini CLI and AI agents to interact with Google Cloud's Agent Registry using gcloud alpha agent-registry commands.
- Automated Authentication: Helps with
gcloud auth loginand project configuration. - Resource Management: Create, list, describe, and delete Agents, MCP Servers, and Endpoints.
- Smart Approvals: Executes read-only commands (
list,describe) immediately, while requesting approval for mutating commands (create,update,delete). - Rich Examples: Built-in support for registering common resources like GitHub MCP Servers, Salesforce Agents, and Vertex AI Model Garden endpoints.
- Python ADK Integration: Specialized support and snippets for the Google Agent Development Kit (ADK).
- GKE Based Agents: Support for configuring Kubernetes Deployments to act as Agents or MCP Servers.
You can install the agent-registry-skill using the GitHub CLI (gh):
gh skill install srinandan/agent-registry-skillOnce installed, you can talk to Gemini in natural language:
- "List my agents in us-central1"
- "Describe the MCP server named github-mcp"
- "Register a new Salesforce agent at https://api.salesforce.com/v1"
- "Show me all registered endpoints"
- "Delete the service called old-test-service"
- "Show me all the agents where the runtime is reasoningEngine"
- "Show agents with identity matching 'my-service-sa'"
- "Which MCP Server has a tool named search_documents?"
- "List agents based on reasoning engine in us-central1"
- Google Cloud SDK version 565.0.0 or higher (minimum).
gcloud alphacomponent version 2026.04.10 or higher (minimum).- Google ADK version 1.31.0 or higher (minimum) for Python integration.
- Proper permissions to access Agent Registry in your Google Cloud project.
The following IAM roles are required to interact with the Agent Registry:
| Role | Access Level |
|---|---|
roles/agentregistry.admin |
Full administrative access |
roles/agentregistry.editor |
Editor access |
roles/agentregistry.viewer |
Read only access |
Please see CONTRIBUTING.md for details on how to contribute to this project.
This demo is NOT endorsed by Google or Google Cloud. The repo is intended for educational/hobbyists use only.
This project is licensed under the terms of the LICENSE file