Skip to main content
hc
Submit
mcpSource-backedReview first Safety ✓ Privacy ✓

PAL MCP Server

Provider Abstraction Layer MCP server for orchestrating multiple AI models, external AI CLIs, planning, consensus, code review, debugging, and delegated sub-agent workflows from one MCP client.

by Beehive Innovations·added 2026-06-05·
Claude CodeClaude Desktop
HarnessClaude CodeClaude Desktop
Review first — review before installing

Open the source and read safety notes before installing.

Safety notes

  • PAL can send code, prompts, files, findings, and conversation context to multiple external model providers.
  • CLI subagents may inspect the workspace, run tools, and return results from separate contexts; restrict roles and provider access.
  • Keep disabled tools disabled unless the workflow needs them, and review tool configuration before broad code analysis or security review.
  • Multi-model consensus can still be wrong; use it as review input, not automatic approval.

Privacy notes

  • Provider API keys, OpenRouter keys, Azure credentials, local model endpoints, and CLI auth tokens are sensitive secrets.
  • Cross-model conversations can reveal proprietary code, customer data, vulnerability details, business plans, and private prompts to several providers.
  • Conversation continuity means earlier context may be forwarded into later tool calls or subagent handoffs.

Prerequisites

  • Python 3.10 or newer, Git, and uv or uvx.
  • API keys for at least one configured provider such as Gemini, OpenAI, OpenRouter, Azure OpenAI, xAI, DIAL, or a local Ollama setup.
  • MCP client such as Claude Code, Codex CLI, Gemini CLI, Cursor, or another compatible host.
  • Clear policy for what code, prompts, screenshots, and context may be shared with external model providers or CLI subagents.

Schema details

Install type
cli
Troubleshooting
No
Source repository stats
Scope
Source repo
Collection metadata
Estimated setup
15 minutes
Difficulty
advanced
Full copyable content
{
  "mcpServers": {
    "pal": {
      "command": "bash",
      "args": [
        "-c",
        "uvx --from git+https://github.com/BeehiveInnovations/pal-mcp-server.git pal-mcp-server"
      ],
      "env": {
        "DEFAULT_MODEL": "auto",
        "GEMINI_API_KEY": "YOUR_GEMINI_API_KEY",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
      }
    }
  }
}

About this resource

Content

PAL MCP is a Provider Abstraction Layer MCP server for coordinating multiple AI models and external AI CLIs from one MCP client. It lets a primary agent ask other providers or CLI subagents for planning, debugging, code review, consensus, pre-commit review, and implementation handoff support.

The project was formerly known as Zen MCP and now documents PAL MCP plus a clink tool for connecting external CLIs such as Gemini CLI, Codex CLI, and Claude Code into a workflow.

Source Review

These sources were reviewed on 2026-06-05. Prefer the live repository for current provider names, model defaults, tool availability, setup scripts, and configuration flags.

Features

  • Multi-model chat, planning, consensus, debugging, and code-review workflows.
  • clink bridge for delegating to external AI CLI subagents.
  • Conversation continuity across PAL tools and model handoffs.
  • Provider support for Gemini, OpenAI, Anthropic, xAI, Azure OpenAI, OpenRouter, DIAL, Ollama, and local models when configured.
  • Optional tool disabling and model defaults through environment configuration.
  • Support for Claude Code, Codex CLI, Gemini CLI, Cursor, and other MCP clients.

Installation

Clone and run the automatic setup:

git clone https://github.com/BeehiveInnovations/pal-mcp-server.git
cd pal-mcp-server
./run-server.sh

The repository also documents an uvx setup shape for MCP clients:

{
  "mcpServers": {
    "pal": {
      "command": "bash",
      "args": [
        "-c",
        "uvx --from git+https://github.com/BeehiveInnovations/pal-mcp-server.git pal-mcp-server"
      ],
      "env": {
        "DEFAULT_MODEL": "auto",
        "GEMINI_API_KEY": "YOUR_GEMINI_API_KEY",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
      }
    }
  }
}

Configure only the provider keys and tools needed for the workflow.

Use Cases

  • Ask a second model to review an implementation plan.
  • Run multi-model consensus before a risky architectural decision.
  • Delegate a security review or bug hunt to a CLI subagent.
  • Continue a discussion through another provider after context compaction.
  • Compare local model feedback against hosted model feedback.

Safety and Privacy

PAL is powerful because it can route context to many providers and CLI subagents. Treat every enabled provider as a data-sharing destination. Use minimal provider keys, limit tools, and avoid forwarding proprietary code, customer data, or vulnerability details unless the provider and workflow are approved.

Duplicate Check

content/mcp/paypal-mcp-server.mdx is unrelated despite the similar text substring. No BeehiveInnovations/pal-mcp-server entry or source URL was found in content/mcp.

#multi-model#cli#planning#code-review#orchestration

Source citations

Signals

Loading live community signals…

More like this, weekly

A short, calm digest of reviewed Claude resources. Unsubscribe any time.