How to Integrate the TestDino MCP Server for Playwright Test Analysis

The TestDino MCP server enables Claude Code and other MCP-compatible agents to audit Playwright test code remotely by connecting to https://mcp.testdino.com, invoking audit tools like health and get_audit_report, and submitting scored markdown reports without storing secrets locally.

Integrating the TestDino MCP server for Playwright test analysis allows AI agents to perform deep, remote audits on your automated testing suites. According to the anthropics/claude-plugins-community repository, this integration leverages a hosted HTTPS endpoint to analyze Playwright specs while keeping authentication credentials secure through OAuth 2.1. The setup requires configuring an MCP client, installing the TestDino plugin, and executing a structured audit flow that scores test quality.

Architecture Overview

The TestDino integration consists of four core components that work together to analyze Playwright tests without executing them locally.

Remote MCP Server – A hosted HTTPS endpoint at https://mcp.testdino.com exposes the TestDino toolset, including health, get_audit_report, and submit_audit_report. This server handles all audit processing remotely, ensuring no test secrets reside in local repositories.

Claude Code Plugin – Located in .claude-plugin/plugin.json, this plugin package registers TestDino skills and routes the agent to the remote MCP server. It acts as the bridge between your local environment and the hosted audit services.

Playwright Test Suite – Your existing Playwright specs serve as the audit target. The TestDino connector routes Playwright code through the audit flow defined in skills/testdino-audit/SKILL.md, analyzing files like tests/login.spec.ts without requiring local test execution.

MCP-Enabled Agent – Claude Code (or any MCP-compatible client) reads the MCP configuration from .mcp.json, loads the TestDino tools, and orchestrates the audit sequence. The agent manages the full lifecycle from health checks to final report submission.

Configuration and Setup

Before running audits, you must configure your MCP client to recognize the TestDino server and install the corresponding plugin.

Add the TestDino MCP server definition to your .mcp.json configuration file:

{
  "mcpServers": {
    "testdino": {
      "type": "http",
      "url": "https://mcp.testdino.com"
    }
  }
}

Source: .mcp.json in the anthropics/claude-plugins-community repository.

Install the TestDino plugin in Claude Code using the marketplace commands:

/plugin marketplace add testdino-hq/TestDino-Plugins
/plugin install testdino@testdino-plugins

Source: README.md installation instructions.

Authentication occurs via OAuth 2.1 when you sign into TestDino, meaning no API keys or secrets are stored in your repository.

Running a Playwright Test Audit

When you request a TestDino audit on Playwright code, the skill executes a specific five-step flow using the MCP tools:

1. Health Check – Call the health tool to verify connectivity and retrieve available projects if the projectId is unknown.

2. Fetch Audit Context – Retrieve read-only context using get_audit_report with action:"context" (or the legacy test_audit tool with action:"analyze"). This provides the background data needed for the audit.

3. Local Review – The agent reviews your local Playwright files and writes findings to a markdown file (e.g., TEST-AUDIT.md). This step involves no remote calls.

4. Submit Report – Upload the completed analysis using submit_audit_report with parameters for projectId, score, and markdownReport. The legacy path uses test_audit with action:"analyze" including score and markdownReport fields.

5. Browse History – Optionally retrieve previous audits using get_audit_report with action:"list" or action:"get" to compare historical scores.

All communication occurs over HTTPS to the remote MCP server. Here is a complete example of the audit workflow in Claude Code:

# Step 1: Verify connectivity and list accessible projects

Use the TestDino connector. Call health and tell me which projects I have access to.

# Step 2: After obtaining a projectId, fetch audit context

> get_audit_report {"projectId":"<your-project-id>", "action":"context"}

# Step 3: Review local files and write findings to TEST-AUDIT.md (local operation)

# Step 4: Submit the scored report

> submit_audit_report {"projectId":"<your-project-id>", "score":8, "markdownReport":"$(cat TEST-AUDIT.md)"}

Source: skills/testdino-audit/SKILL.md.

You can also list previous audits programmatically:

# List all previous audits for a project

audit_list = get_audit_report({"projectId": "<your-project-id>", "action": "list"})
print(audit_list)

Key Files and Implementation Details

Understanding the source files helps troubleshoot integration issues:

• testdino/README.md – Contains overview, quick-start instructions, and compatibility details for OAuth 2.1 authentication.
• testdino/.mcp.json – Defines the MCP server connection settings used by Claude Code and other agents.
• testdino/.claude-plugin/plugin.json – Stores plugin metadata that registers TestDino skills with the agent.
• testdino/skills/testdino-audit/SKILL.md – Specifies the complete audit flow, tool selection logic, and rules for scoring Playwright tests.

Summary

• Configure the MCP client by adding the TestDino server definition to .mcp.json with the endpoint https://mcp.testdino.com.
• Install the plugin via Claude Code marketplace commands to enable skill registration.
• Execute the audit flow using health, get_audit_report, and submit_audit_report tools in sequence.
• Store no secrets locally; authentication uses OAuth 2.1 through the remote MCP server.
• Reference SKILL.md for the authoritative specification of audit steps and tool parameters.

Frequently Asked Questions

What MCP tools does TestDino expose for Playwright analysis?

TestDino exposes health, get_audit_report, and submit_audit_report as primary tools. Legacy implementations may reference test_audit with different action parameters. These tools handle connectivity checks, context retrieval, and report submission over HTTPS.

How does authentication work with the TestDino MCP server?

Authentication uses OAuth 2.1 when you sign into TestDino through the browser. No API keys or credentials are stored in your local repository or .mcp.json file. The remote server manages all authentication states, ensuring secrets remain secure.

Can I use TestDino with MCP clients other than Claude Code?

Yes. Any MCP-compatible agent that supports the HTTP transport type defined in .mcp.json can load the TestDino tools. The plugin in .claude-plugin/plugin.json is specific to Claude Code, but the underlying MCP server at https://mcp.testdino.com follows the standard Model Context Protocol.

What file should I check for the complete audit workflow specification?

The authoritative source for the audit workflow is skills/testdino-audit/SKILL.md. This file details the exact sequence of tool calls, parameter formats for submit_audit_report, and rules for generating the markdown report that accompanies your Playwright test analysis.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}