feat: Add "secret" GitHub authentication with wrapper scripts

[jfi]

Overview

Following on from the discussion at #6, this PR enhances the GitHub MCP Extension for Zed by implementing automatic token authentication through wrapper scripts, addressing the limitation of Zed's WASM sandbox which cannot access host environment variables.

Problem Solved

Previously, users had to manually obtain and configure GitHub personal access tokens. This enhancement allows the extension to automatically use existing GitHub authentication from:

• GitHub CLI (gh auth status)
• Environment variables (GITHUB_TOKEN)
• Token files (.github_token)

Key Features

🔄 Automatic Authentication

• Auto-detection: Extension automatically detects and uses wrapper scripts
• Multi-method support: GitHub CLI, environment variables, and token files
• Cross-platform: Wrapper scripts for JavaScript (Node.js), Bash, and PowerShell
• Backwards compatibility: Falls back to manual token configuration

📁 Improved Project Structure

• Organized documentation: Dedicated configuration and troubleshooting guides
• Wrapper scripts: Self-contained authentication handlers in wrappers/ directory
• Clean configuration: Simplified settings with optional wrapper script usage

🛠 Enhanced User Experience

• One-click setup: Simply set "use_wrapper_script": true in settings
• Better error handling: Clear feedback when authentication fails
• Comprehensive docs: Step-by-step guides for all supported platforms

NOTE: This should support Windows, but I don't have a machine lying around to test it.

Technical Implementation

Wrapper Script Auto-Detection

The extension automatically searches for wrapper scripts in the following order:

1. github-mcp-wrapper.js (Node.js)
2. github-mcp-wrapper.sh (Bash)
3. github-mcp-wrapper.ps1 (PowerShell)

Authentication Flow

flowchart TD
    A[Extension Start] --> B{use_wrapper_script?}
    B -->|true| C[Detect Wrapper Script]
    B -->|false| D[Use Manual Token]
    C --> E{Script Found?}
    E -->|yes| F[Execute Wrapper]
    E -->|no| G[Log Warning + Use Manual Token]
    F --> H{Auth Successful?}
    H -->|yes| I[Initialize MCP Server]
    H -->|no| J[Show Error]

Loading

Configuration

Simple Setup

{
  "context_servers": {
    "mcp-server-github": {
      "source": "extension",
      "settings": {
        "use_wrapper_script": true
      }
    }
  }
}

Manual Fallback (Unchanged)

{
  "context_servers": {
    "mcp-server-github": {
      "source": "extension",
      "settings": {
        "github_token": "your_token_here"
      }
    }
  }
}

Files Added/Modified

New Files

• wrappers/github-mcp-wrapper.js - Node.js authentication wrapper
• wrappers/github-mcp-wrapper.sh - Bash authentication wrapper
• wrappers/github-mcp-wrapper.ps1 - PowerShell authentication wrapper
• wrappers/README.md - Wrapper script documentation
• docs/configuration.md - Comprehensive configuration guide
• docs/troubleshooting.md - Common issues and solutions
• CHANGELOG.md - Version history

Modified Files

• src/mcp_server_github.rs - Core extension logic with wrapper support
• README.md - Updated with new features and simplified setup
• configuration/default_settings.jsonc - Added wrapper script option

Removed Files

• configuration/installation_instructions.md - Consolidated into docs/

Testing

The enhancement has been tested with:

• ✅ GitHub CLI authentication (gh auth login)
• ✅ Environment variable authentication (GITHUB_TOKEN)
• ✅ Token file authentication (.github_token)
• ✅ Manual token fallback
• ✅ Error handling for missing/invalid authentication

Backwards Compatibility

• ✅ All existing configurations continue to work unchanged
• ✅ Manual token configuration remains fully supported
• ✅ No breaking changes to existing functionality
• ✅ Graceful fallback when wrapper scripts are unavailable

[jeffreyguenther]

I appreciate all the work that went into this. I prefer a first-party solution. The Zed team is keen to improve the extensions API, so I've taken this up with them.

Correct me if I'm wrong. It's my understanding the key thing we need is to have access the shell env of the user's machine. Then we can access the correct environment variables.