Skip to main content

AiDD MCP Server

IMPORTANT: This package has been renamed to skydeckai-code and is now maintained at https://pypi.org/project/skydeckai-code/. All new updates and improvements will be added to the new package. This mcp-server-aidd package is no longer maintained.

smithery badge

An MCP server that provides a comprehensive set of tools for AI-driven development workflows. Features include file system operations, code analysis using tree-sitter for multiple programming languages, Git operations, code execution, and system information retrieval. Designed to enhance AI's capability to assist in software development tasks.

AiDD Server MCP server

Installation

Installing via Smithery

To install AiDD for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install mcp-server-aidd --client claude

Manual Installation

# Using mcp-get
npx @michaellatman/mcp-get@latest install mcp-server-aidd

# Using pip
pip install mcp-server-aidd

# Using uv
uvx mcp-server-aidd

Claude Desktop Setup

Add to your claude_desktop_config.json:

{
    "mcpServers": {
        "aidd-ai-software-development-utilities": {
            "command": "uvx",
            "args": ["mcp-server-aidd"]
        }
    }
}

SkyDeck AI Helper App

If you're using SkyDeck AI Helper app, you can search for "AiDD" and install the mcp-server-aidd.

SkyDeck AI Helper App

Key Features

  • File system operations (read, write, edit, move, delete)
  • Directory management and traversal
  • Multi-language code analysis using tree-sitter
  • Multi-language code execution with safety measures
  • Git operations (status, diff, commit, branch management)
  • Security controls with configurable workspace boundaries
  • Screenshot and screen context tools
  • Image handling tools

Available Tools

Basic File Operations

Tool Parameters Returns
read_file path: string File content
read_multiple_files paths: string[] Multiple file contents with headers
write_file path: string, content: string Success confirmation
move_file source: string, destination: string Success confirmation
delete_file path: string Success confirmation
get_file_info path: string File metadata (size, timestamps, permissions)

Common usage:

# Read file
aidd-cli --tool read_file --args '{"path": "src/main.py"}'

# Write file
aidd-cli --tool write_file --args '{"path": "output.txt", "content": "Hello World"}'

# Get file info
aidd-cli --tool get_file_info --args '{"path": "src/main.py"}'

Complex File Operations

edit_file

Pattern-based file editing with preview support:

{
    "path": "src/main.py",
    "edits": [
        {
            "oldText": "def old_function():",
            "newText": "def new_function():"
        }
    ],
    "dryRun": false,
    "options": {
        "preserveIndentation": true,
        "normalizeWhitespace": true,
        "partialMatch": true
    }
}

Returns: Diff of changes or preview in dry run mode.

Directory Operations

Tool Parameters Returns
get_allowed_directory none Current allowed directory path
update_allowed_directory directory: string (absolute path) Success confirmation
list_directory path: string Directory contents list
create_directory path: string Success confirmation
search_files pattern: string, path?: string, include_hidden?: boolean Matching files list

directory_tree

Generates complete directory structure:

{
    "path": "src",
    "include_hidden": false
}

Returns: JSON tree structure of directory contents.

Common usage:

# List directory
aidd-cli --tool list_directory --args '{"path": "."}'

# Search for Python files
aidd-cli --tool search_files --args '{"pattern": ".py", "path": "src"}'

Git Operations

Tool Parameters Returns
git_init path: string, initial_branch?: string Repository initialization status
git_status repo_path: string Working directory status
git_add repo_path: string, files: string[] Staging confirmation
git_reset repo_path: string Unstaging confirmation
git_checkout repo_path: string, branch_name: string Branch switch confirmation

Complex Git Operations

git_commit
{
    "repo_path": ".",
    "message": "feat: add new feature"
}

Returns: Commit hash and confirmation.

git_diff
{
    "repo_path": ".",
    "target": "main"
}

Returns: Detailed diff output.

git_log
{
    "repo_path": ".",
    "max_count": 10
}

Returns: Array of commit entries with hash, author, date, and message.

Common usage:

# Check status
aidd-cli --tool git_status --args '{"repo_path": "."}'

# Create and switch to new branch
aidd-cli --tool git_create_branch --args '{"repo_path": ".", "branch_name": "feature/new-branch"}'

Code Analysis

tree_sitter_map

Analyzes source code structure:

{
    "path": "src"
}

Returns:

  • Classes and their methods
  • Functions and parameters
  • Module structure
  • Code organization statistics
  • Inheritance relationships

Supported Languages:

  • Python (.py)
  • JavaScript (.js, .jsx, .mjs, .cjs)
  • TypeScript (.ts, .tsx)
  • Java (.java)
  • C++ (.cpp, .hpp, .cc)
  • Ruby (.rb, .rake)
  • Go (.go)
  • Rust (.rs)
  • PHP (.php)
  • C# (.cs)
  • Kotlin (.kt, .kts)

System Information

Tool Parameters Returns
get_system_info none Comprehensive system details

Returns:

{
  "working_directory": "/path/to/project",
  "system": {
    "os", "os_version", "architecture", "python_version"
  },
  "wifi_network": "MyWiFi",
  "cpu": {
    "physical_cores", "logical_cores", "total_cpu_usage"
  },
  "memory": { "total", "available", "used_percentage" },
  "disk": { "total", "free", "used_percentage" },
  "mac_details": {  // Only present on macOS
    "model": "Mac mini",
    "chip": "Apple M2",
    "serial_number": "XXX"
  }
}

Provides essential system information in a clean, readable format.

# Get system information
aidd-cli --tool get_system_info

Screen Context and Image Tools

get_active_apps

Returns a list of currently active applications on the user's system.

{
    "with_details": true
}

Parameters:

Parameter Type Required Description
with_details boolean No Whether to include additional details about each application (default: false)

Returns:

{
    "success": true,
    "platform": "macos",
    "app_count": 12,
    "apps": [
        {
            "name": "Firefox",
            "has_windows": true,
            "window_count": 3,
            "visible_windows": [
                { "name": "GitHub - Mozilla Firefox", "width": 1200, "height": 800 }
            ]
        },
        {
            "name": "VSCode",
            "has_windows": true
        }
    ]
}

This tool provides valuable context about applications currently running on the user's system, which can help with providing more relevant assistance.

get_available_windows

Returns detailed information about all available windows currently displayed on the user's screen.

{}

Returns:

{
    "success": true,
    "platform": "macos",
    "count": 8,
    "windows": [
        {
            "id": 42,
            "title": "Document.txt - Notepad",
            "app": "Notepad",
            "visible": true
        },
        {
            "title": "Terminal",
            "app": "Terminal",
            "visible": true,
            "active": true
        }
    ]
}

This tool helps understand what's visible on the user's screen and can be used for context-aware assistance.

capture_screenshot

Captures a screenshot of the user's screen or a specific window.

{
    "output_path": "screenshots/capture.png",
    "capture_mode": {
        "type": "named_window",
        "window_name": "Visual Studio Code"
    }
}

Parameters:

Parameter Type Required Description
output_path string No Path where the screenshot should be saved (default: generated path)
capture_mode object No Specifies what to capture
capture_mode.type string No Type of screenshot: 'full', 'active_window', or 'named_window' (default: 'full')
capture_mode.window_name string No Name of window to capture (required when type is 'named_window')

Returns:

{
    "success": true,
    "path": "/path/to/screenshots/capture.png"
}

This tool captures screenshots for visualization, debugging, or context-aware assistance.

read_image_file

Reads an image file from the file system and returns its contents as a base64-encoded string.

{
    "path": "images/logo.png"
}

Parameters:

Parameter Type Required Description
path string Yes Path to the image file to read
max_size integer No Maximum file size in bytes (default: 100MB)

Returns: Base64-encoded image data that can be displayed or processed.

This tool supports common image formats like PNG, JPEG, GIF, and WebP, and automatically resizes images for optimal viewing.

Code Execution

execute_code

Executes code in various programming languages with safety measures and restrictions.

{
    "language": "python",
    "code": "print('Hello, World!')",
    "timeout": 5
}

Supported Languages:

  • Python (python3)
  • JavaScript (Node.js)
  • Ruby
  • PHP
  • Go
  • Rust

Parameters:

Parameter Type Required Description
language string Yes Programming language to use
code string Yes Code to execute
timeout integer No Maximum execution time (default: 5s)

Example Usage:

# Python example
aidd-cli --tool execute_code --args '{
    "language": "python",
    "code": "print(sum(range(10)))"
}'

# JavaScript example
aidd-cli --tool execute_code --args '{
    "language": "javascript",
    "code": "console.log(Array.from({length: 5}, (_, i) => i*2))"
}'

# Ruby example
aidd-cli --tool execute_code --args '{
    "language": "ruby",
    "code": "puts (1..5).reduce(:+)"
}'

# Go example
aidd-cli --tool execute_code --args '{
    "language": "go",
    "code": "fmt.Println(\"Hello, Go!\")"
}'

Requirements:

  • Respective language runtimes must be installed
  • Commands must be available in system PATH
  • Proper permissions for temporary file creation

⚠️ Security Warning: This tool executes arbitrary code on your system. Always:

  1. Review code thoroughly before execution
  2. Understand the code's purpose and expected outcome
  3. Never execute untrusted code
  4. Be aware of potential system impacts
  5. Monitor execution output

execute_shell_script

Executes shell scripts (bash/sh) with safety measures and restrictions.

{
    "script": "echo \"Current directory:\" && pwd",
    "timeout": 300
}

Parameters:

Parameter Type Required Description
script string Yes Shell script to execute
timeout integer No Maximum execution time (default: 300s, max: 600s)

Example Usage:

# List directory contents with details
aidd-cli --tool execute_shell_script --args '{
    "script": "ls -la"
}'

# Find all Python files recursively
aidd-cli --tool execute_shell_script --args '{
    "script": "find . -name \"*.py\" -type f"
}'

# Complex script with multiple commands
aidd-cli --tool execute_shell_script --args '{
    "script": "echo \"System Info:\" && uname -a && echo \"\nDisk Usage:\" && df -h"
}'

Features:

  • Uses /bin/sh for maximum compatibility across systems
  • Executes within the allowed directory
  • Separate stdout and stderr output
  • Proper error handling and timeout controls

⚠️ Security Warning: This tool executes arbitrary shell commands on your system. Always:

  1. Review the script thoroughly before execution
  2. Understand the script's purpose and expected outcome
  3. Never execute untrusted scripts
  4. Be aware of potential system impacts
  5. Monitor execution output

Configuration

Configuration file: ~/.aidd/config.json

{
    "allowed_directory": "/path/to/workspace"
}

CLI Usage

Basic command structure:

aidd-cli --tool <tool_name> --args '<json_arguments>'

# List available tools
aidd-cli --list-tools

# Enable debug output
aidd-cli --debug --tool <tool_name> --args '<json_arguments>'

Debugging

Use MCP Inspector for debugging:

npx @modelcontextprotocol/inspector run

Security

  • Operations restricted to configured allowed directory
  • Path traversal prevention
  • File permission preservation
  • Safe operation handling

Upcoming Features

  • GitHub tools:
    • PR Description Generator
    • Code Review
    • Actions Manager
  • Pivotal Tracker tools:
    • Story Generator
    • Story Manager

Development Status

Currently in active development. Features and API may change.

License

Apache License 2.0 - see LICENSE

Star History Chart

Metadata

Release files for mcp-server-aidd 0.1.23

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for mcp-server-aidd 0.1.23
File Size Uploaded Provenance
mcp_server_aidd-0.1.23.tar.gz 176.7 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for mcp-server-aidd 0.1.23
File Interpreter ABI Platform
mcp_server_aidd-0.1.23-py3-none-any.whl Python 3 none any Details

Total release size: 235.6 kB

Release files / mcp_server_aidd-0.1.23.tar.gz

Download URL mcp_server_aidd-0.1.23.tar.gz
Size 176.7 kB
Tags Source
SHA-256 checksum
How to use checksums
186f3a96c23d6b7a629ad249a2bba00d0afe27e301b3594f871f162c87c2f7f4
BLAKE2b-256 checksum
How to use checksums
6e9ff98840bab0c50659ec5275b6715f19bdc361267c464286d5a785c147ee41
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.13.0

Release files / mcp_server_aidd-0.1.23-py3-none-any.whl

Download URL mcp_server_aidd-0.1.23-py3-none-any.whl
Size 58.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
bde8fe0c3b5051bb51931f9b53352a053b00ceecb5e90260e6e932c91d23b960
BLAKE2b-256 checksum
How to use checksums
3f812e836b03abf292a21b785ce4d3c2f825b0d47c8165433ed899141584c0a6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.13.0

Security for mcp-server-aidd 0.1.23

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page