Skip to content

Other LLM Integrations

Setup guides for Gemini CLI and Codex CLI - additional platforms with napari MCP support.

🌐 Gemini CLI

Setup guide for Google's Gemini CLI with napari MCP server.

Quick Setup

# 1. Install napari-mcp
pip install napari-mcp

# 2. Auto-configure Gemini CLI (global)
napari-mcp-install gemini --global

# OR configure for specific project
napari-mcp-install gemini --project /path/to/project

Configuration Locations

  • Global: ~/.gemini/settings.json
  • Project: .gemini/settings.json in project root

Installation Options

# Global installation (recommended)
napari-mcp-install gemini --global

# Project-specific installation
napari-mcp-install gemini --project .

# Preview changes
napari-mcp-install gemini --dry-run

# Use persistent Python
napari-mcp-install gemini --persistent

Manual Configuration

Gemini CLI supports additional configuration options:

{
  "mcpServers": {
    "napari-mcp": {
      "command": "uv",
      "args": ["run", "--with", "napari-mcp", "napari-mcp"],
      "cwd": ".",
      "timeout": 600000,
      "trust": false
    }
  }
}

Configuration fields: - cwd: Working directory for the server - timeout: Timeout in milliseconds (default: 600000 = 10 minutes) - trust: If true, bypasses tool confirmation prompts (use with caution)

Management

# Check installation
napari-mcp-install list

# Update configuration
napari-mcp-install gemini --global --force

# Uninstall
napari-mcp-install uninstall gemini

Troubleshooting

Gemini doesn't see napari tools

  1. Verify installation: napari-mcp-install list
  2. Check config file exists: cat ~/.gemini/settings.json
  3. Restart Gemini CLI
  4. Reinstall: napari-mcp-install gemini --global --force

🤖 Codex CLI

Setup guide for OpenAI's Codex CLI with napari MCP server.

Quick Setup

# 1. Install napari-mcp
pip install napari-mcp

# 2. Auto-configure Codex CLI
napari-mcp-install codex

# 3. Restart Codex CLI

Configuration Location

  • All platforms: ~/.codex/config.toml

TOML Format

Codex CLI uses TOML configuration format instead of JSON.

Installation Options

# Basic installation
napari-mcp-install codex

# Preview changes
napari-mcp-install codex --dry-run

# Use persistent Python
napari-mcp-install codex --persistent

# Force update
napari-mcp-install codex --force

Manual Configuration

Edit ~/.codex/config.toml:

[mcp_servers.napari_mcp]
command = "uv"
args = ["run", "--with", "napari-mcp", "napari-mcp"]

Using Persistent Environment

[mcp_servers.napari_mcp]
command = "python"
args = ["-m", "napari_mcp.server"]

Management

# Check installation
napari-mcp-install list

# Update configuration
napari-mcp-install codex --force

# Uninstall
napari-mcp-install uninstall codex

Troubleshooting

Codex doesn't see napari tools

  1. Verify installation: napari-mcp-install list
  2. Check config file: cat ~/.codex/config.toml
  3. Validate TOML syntax: python -c "import toml; toml.load(open('.codex/config.toml'))"
  4. Restart Codex CLI
  5. Reinstall: napari-mcp-install codex --force

TOML syntax errors

The installer requires the toml package. Install it: 

pip install toml

🔄 Install for All Platforms

You can install napari-mcp for all supported platforms at once:

napari-mcp-install all

This installs for: - Claude Desktop - Claude Code - Cursor (global) - Cline (VS Code and Cursor) - Gemini (global) - Codex

Use --dry-run to preview: 

napari-mcp-install all --dry-run

⚖️ Platform Comparison

Feature Gemini CLI Codex CLI
Config Format JSON TOML
Project-specific ✅ Yes ❌ No
Global install ✅ Yes ✅ Yes
Tool confirmations Configurable (trust) Yes
Timeout setting ✅ Configurable Default
maturity Beta Stable

🔐 Security Notes

Gemini CLI

  • trust: true bypasses all tool confirmation prompts
  • Only set trust: true in controlled environments
  • Use trust: false (default) for safety

Codex CLI

  • Always requires confirmation for code execution
  • Runs with your user permissions
  • TOML config file should have appropriate permissions

📚 Resources

Gemini CLI

Codex CLI

📚 Next Steps

→ Back to Integrations Overview