Suggestion: Clarvia AEO integration for MCP tool quality validation

[digitamaz]

Clarvia AEO Integration for ADK Tool Quality

Hi Google ADK team!

I wanted to share Clarvia — an AEO (AI Engine Optimization) quality scanner for MCP servers and agent tools — as a potential integration or example for ADK users.

The problem Clarvia solves

When ADK agents use MCP tools, tool quality varies significantly. Tools with weak descriptions cause agents to misuse them or fail to invoke them at all. There's currently no standard way to assess "is this MCP tool ready for agent use?"

What Clarvia provides

Clarvia indexes and scores 27,906+ MCP servers, APIs, CLIs, and Skills across 24 quality dimensions:

• Description clarity — can the model understand when to call this tool?
• Parameter documentation — are all params clearly typed and described?
• Response format consistency — does the tool return predictable output?
• Error handling quality — does it fail gracefully?
• AEO aggregate score — single grade (A+ to F)

ADK integration example

from google.adk.agents import Agent
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset, StdioServerParameters

# Check AEO quality before adding to agent
# GET https://clarvia-api.onrender.com/v1/tools/{server}/aeo
# → {"aeo_score": 87, "grade": "A", "ready_for_agents": true}

agent = Agent(
    model='gemini-2.0-flash',
    tools=[MCPToolset(StdioServerParameters(command='my-quality-checked-mcp-server'))]
)

Resources

• 🌐 Platform: https://clarvia.art
• 📦 MCP server: npx clarvia-mcp-server (1,490 weekly downloads, How to use api_transport "REST"? in Agent #1 npm for 'mcp quality scanner')
• 📖 API: https://clarvia-api.onrender.com/docs (132 endpoints, OpenAPI 3.1)
• 🐙 GitHub: https://github.com/clarvia-project/scanner (MIT)

Would be happy to create an example showing Clarvia-gated tool selection in an ADK agent workflow. Thanks for the great work on ADK!