Clarvia MCP: Tool quality scoring for DeepEval benchmarking pipelines

[digitamaz]

Clarvia + DeepEval: Scoring the tools in your eval stack

Hi DeepEval team,

We built Clarvia — an MCP server and API that provides Agent Engine Optimization (AEO) scoring for AI tools, APIs, and services.

Why relevant to evaluation frameworks

When building LLM evaluation pipelines, the quality of tools your agents use directly impacts eval results. Clarvia can score:

• Tool documentation completeness (are instructions clear for LLMs?)
• API schema quality (are inputs/outputs well-typed?)
• Agent-readability of tool descriptions
• Overall "agent-usability" score

Integration with DeepEval

from deepeval import evaluate
from deepeval.metrics import ToolCorrectnessMetric

# Pre-evaluate tool quality before including in test suite
import requests
tool_quality = requests.get(
    "https://clarvia-api.onrender.com/v1/scan",
    params={"url": "https://your-tool-api.com"}
).json()

Resources

• API: https://clarvia-api.onrender.com/docs
• MCP Server: npx clarvia-mcp-server (24 tools)
• npm: https://www.npmjs.com/package/clarvia-mcp-server (714 downloads/week)
• Platform: https://clarvia.art (15,400+ indexed tools with AEO scores)

—The Clarvia team