mcp-client

Connect to MCP servers and call tools with structured arguments.

Updated Jan 28, 2026
One-click install
npx skills add https://github.com/cristian-robert/autonomous-automation-tester --skill mcp-client-cristian-robert
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcp-client
Source: https://github.com/cristian-robert/autonomous-automation-tester/tree/main/.claude/skills/mcp-client
Command: npx skills add https://github.com/cristian-robert/autonomous-automation-tester --skill mcp-client-cristian-robert

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a universal MCP client that connects to any MCP server on demand, wrapping MCP tool schemas to avoid context window bloat and enabling on-demand tool discovery, execution, and integration with external MCP servers (Playwright, GitHub, filesystem, etc.).

Core Features & Use Cases

  • On-demand tool discovery: list available tools from configured MCP servers without loading full schemas.
  • Flexible execution: call tools with structured JSON arguments and receive parsed results.
  • Multi-server integration: connect to Playwright, GitHub, filesystem, and other MCP servers, orchestrating workflows from a single client.
  • Use Case: Design an automated QA workflow that discovers tests on a Playwright MCP server, runs selected tools, and aggregates results for reporting.

Quick Start

  • Install required runtime and MCP libraries: pip install mcp fastmcp
  • See available servers: python scripts/mcp_client.py servers
  • List tools for a server (e.g., playwright): python scripts/mcp_client.py tools playwright
  • Call a simple tool to navigate a page: python scripts/mcp_client.py call playwright browser_navigate '{"url": "https://example.com"}'
  • For multi-step interactions, use browser_run_code to keep the same browser session: python scripts/mcp_client.py call playwright browser_run_code '{"code": "await page.goto("https://example.com"); return await page.title();"}'

Frequently Asked Questions about mcp-client

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I connect to an MCP server on demand without bloating the context window?

You can connect to an MCP server on demand using a universal MCP client that wraps tool schemas, preventing context bloat while enabling on-demand tool discovery and execution for external servers like Playwright or GitHub.

What is the best way to list and call tools on a Playwright MCP server using Python?

The best way to list and call tools on a Playwright MCP server is using a Python interface that discovers available tools and executes them with structured JSON arguments via configurable transport backends like stdio or streamable_http.

Can I use multiple MCP transports like stdio and SSE with a single client?

Yes, you can use multiple MCP transports with a single client by configuring the MCP config resolver to support stdio, SSE, streamable_http, and fastmcp backends for flexible multi-server integration.

How do I run automated QA workflows across different MCP servers?

You can run automated QA workflows by connecting to a Playwright MCP server, discovering testing tools, executing them with structured arguments, and aggregating the parsed results for reporting from a single Python client interface.

Do I need to install specific libraries to start integrating with external MCP servers?

Yes, you need to install Python runtime environments and MCP libraries using pip install mcp fastmcp before connecting to external MCP servers for tool discovery and execution.

Why does loading full tool schemas cause issues and how is this avoided?

Loading full tool schemas causes context window bloat, which is avoided by using a universal MCP client that wraps the schemas and only loads tool schemas on demand during discovery and execution.