pal-mcp-server

PAL MCP (Provider Abstraction Layer, formerly Zen MCP) is an MCP server that lets the coding CLI you already use, such as Claude Code, Gemini CLI, Codex CLI or Qwen Code, bring other models into the same task. From one prompt you can ask Gemini, OpenAI, Anthropic, Grok, Azure, OpenRouter or local Ollama models for a second opinion, a code review or a consensus decision, while your main CLI stays in control and context carries across tools.

It suits developers who want multi-model workflows without switching tools: a code review by two different models before a commit, a debugging session with systematic root-cause analysis, planning followed by implementation, or a pre-commit check. Its `clink` tool goes further and connects external CLIs as subagents, so Claude Code can spawn an isolated Codex reviewer or Codex can launch a Gemini CLI instance, each with its own role and fresh context.

Getting started: configure API keys for the providers you want to use (or a local Ollama endpoint), add the server to your CLI's MCP configuration following the repository's setup guide, then ask for workflows such as "perform a codereview using gemini pro and o3".