ck:use-mcp

Execute MCP server tool operations with JSON-only results via Gemini CLI.

Updated Mar 7, 2026
One-click install
npx skills add https://github.com/MinhHoangDono/antigravity-kit --skill ck-use-mcp-minhhoangdono
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ck:use-mcp
Source: https://github.com/MinhHoangDono/antigravity-kit/tree/main/.agent/skills/use-mcp
Command: npx skills add https://github.com/MinhHoangDono/antigravity-kit --skill ck-use-mcp-minhhoangdono

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the problem of reliably executing MCP server tools by choosing an execution path that preserves context and returns machine-parseable results.

Core Features & Use Cases

  • Gemini CLI MCP-safe execution: Runs MCP operations via stdin piping so MCP initialization is not skipped, enabling consistent tool usage.
  • Structured, JSON-only outputs: Produces parseable results shaped like server/tool success responses for downstream handling.
  • Fallback to tool-discovery execution: Uses an mcp-manager subagent when Gemini CLI is unavailable, with guidance to repair issues using an existing builder skill.

Quick Start

Tell your AI assistant: use ck:use-mcp to perform the task you want by selecting an available MCP tool and returning JSON output.

Frequently Asked Questions about ck:use-mcp

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

FAQPage Schema
How do I get structured JSON output from MCP server tools?

MCP tools can return structured JSON output by executing operations through stdin piping with auto-loaded JSON formatting, ensuring machine-parseable results for downstream automation workflows.

Why does Gemini CLI skip MCP initialization when running tools?

Gemini CLI skips MCP initialization when using deprecated flags, so you must use stdin piping to preserve context and ensure consistent MCP tool execution and discovery.

What is the fallback strategy when Gemini CLI is unavailable for MCP tool execution?

The fallback strategy for MCP tool execution without Gemini CLI involves using an mcp-manager subagent to discover and run tools, with guidance to repair issues using an existing builder skill.

How do I discover available MCP tools and run them for task automation?

Discovering and running MCP tools for task automation requires deterministic selection logic, executing the chosen tool via stdin piping, and returning JSON-only results to complete downstream workflows.

Do I need a specific environment setup to execute MCP tools via stdin piping?

Executing MCP tools via stdin piping requires the Gemini CLI as the preferred execution path and utilizes auto-loaded JSON formatting via GEMINI.md, avoiding deprecated flags to prevent skipped initialization.