meta-mcp-wizard

What This Skill Does

The meta-mcp-wizard skill provides an interactive bridge between your AI agent and the MCPHero platform. It allows users to build, deploy, and manage persistent Model Context Protocol (MCP) servers directly from within their preferred AI environment, such as Claude Desktop or Cursor. By utilizing a wizard-driven pipeline, this skill abstracts the complexity of creating custom MCP tools. Instead of manual configuration or repeated prompt engineering for recurring tasks, the agent generates a hosted, persistent server instance. This optimization can reduce token consumption by up to 99% for repetitive operations, effectively transforming your transient agent tasks into robust, reusable software components.

Installation

To integrate this skill into your environment, use the OpenClaw command-line interface. Run the following in your terminal:

clawhub install openclaw/skills/skills/arterialist/meta-mcp-creator

After installation, configure your local MCP client (e.g., Claude Desktop) by adding the following server configuration:

{
  "mcpServers": {
    "mcphero": {
      "url": "https://api.mcphero.app/mcp/meta/mcp"
    }
  }
}

Ensure you restart your MCP client after saving the configuration to ensure the Meta-MCP server is correctly initialized.

Use Cases

This skill is designed for developers and power users who need to:

• Automate recurring API interactions without rewriting prompts.
• Create specialized internal tools for data retrieval from private sources.
• Build persistent custom tooling that needs to be shared or maintained across multiple chat sessions.
• Manage complex environmental variables and authentication tokens for external services via a structured wizard interface.

Example Prompts

1. "I need to automate my daily ticket updates in Jira. Can you help me use the Meta-MCP wizard to create a persistent server that interfaces with the Jira API?"
2. "Start a new session with the Meta-MCP wizard. I want to build a tool that fetches my latest repository stats from GitHub and formats them into a markdown report."
3. "After generating my code, run the deploy command for the current Meta-MCP session and give me the server URL so I can connect it to my main project."

Tips & Limitations

• Sync vs. Async: Always remember to call wizard_state after triggering an asynchronous tool like wizard_generate_code to confirm progress.
• Authentication: OAuth 2.1 is handled automatically by the client upon first connection, so there is no need to manually handle complex auth headers.
• State Management: Keep track of your server_id returned by wizard_create_session; it is required for all subsequent interactions in that pipeline session.
• Scope: This tool is specifically for managing MCP servers via the MCPHero infrastructure and is not intended for general-purpose file management outside of the wizard-defined scope.