CCG MCP Tool (Claude Code/Codex/Gemini)

CCG MCP Tool (Claude Code/Codex/Gemini)

This is a powerful Model Context Protocol (MCP) server that integrates multiple AI coding agents—Anthropic Claude Code, OpenAI Codex, and Google Gemini—directly into your workflow. It enables seamless cross-provider analysis, leveraging Gemini's massive token window, Codex's specialized coding capabilities, and Claude's advanced reasoning.

• Ask any supported AI for its perspective.
• Brainstorm ideas with multi-provider frameworks.
• New: Apply research-grounded mitigation skills to avoid common AI coding mistakes.

Goal: Use the best AI for the job. Analyze massive codebases with Gemini, get precision edits with Codex, or use Claude's reasoning—all from a single MCP interface.

Prerequisites

Ensure you have the following CLI tools installed and configured:

1. Node.js (v16.0.0 or higher)
2. Google Gemini CLI
3. Codex CLI (optional)
4. Claude Code (optional)

Installation

claude mcp add ccg-tool -- npx -y ccg-mcp-tool

Configuration

Register the MCP server with your MCP client. You can set the default provider and model via command-line arguments:

For NPX Usage (Recommended)

Add this configuration to your Claude Desktop config file:

{
  "mcpServers": {
    "ccg-tool": {
      "command": "npx",
      "args": [
        "-y", 
        "ccg-mcp-tool",
        "--provider",
        "codex",
        "--model",
        "gpt-5.3-codex"
      ]
    }
  }
}

Note: The --provider and --model startup arguments set the global default for all tool calls in your session. Tools still support overriding these per-request.

Example Workflow

• Multi-Provider: ask ai --provider codex --model gpt-5.3-codex to refactor @src/auth.ts
• Mistake Mitigation: mitigate mistakes --skill requirements-grounding for @new-feature.md
• Gemini Specific: ask ai --provider gemini to analyze @. and explain the architecture

Tools (for the AI)

ask-ai

Universal tool for AI analysis across providers.

• prompt (required): Your request. Use @ for files.
• provider (optional): gemini, codex, or claude. Defaults to server config.
• model (optional): Specific model for the provider (e.g., gemini-2.5-flash, gpt-5.3-codex).
• sandbox (optional): Gemini-only. Run in an isolated environment.
• changeMode (optional): Gemini-only. Returns structured edits.

mitigate-mistakes

Apply research-grounded gates to prevent common AI agent failure modes.

• skill (required): The mitigation gate to apply (e.g., requirements-grounding, secure-coding-and-validation-gate).
• prompt (required): The task or code to analyze.
• provider (optional): Which AI to use for the assessment.

Available Skills:

• requirements-grounding: Prevent requirement-conflicting hallucinations.
• context-scope-discipline: Maintain focus on the requested change set.
• dependency-verification: Verify imports and versions.
• design-doc-and-architecture-gate: Ensure structural integrity.
• test-and-error-path-gate: Validate edge cases and failure modes.
• secure-coding-and-validation-gate: Identify security risks (unbiased).
• code-quality-enforcer: Check for maintainability and patterns.
• deterministic-validation-gate: Use evidence-based checks.
• code-review-and-change-gate: Final gate before application.

brainstorm

Generate ideas using structured methodologies like SCAMPER or Design Thinking.

• methodology: divergent, convergent, scamper, lateral, etc.
• domain: Specialized context (e.g., software, security).

License

MIT License.

Disclaimer: This is an unofficial tool and is not affiliated with Google, OpenAI, or Anthropic.