FastMCP for PowerShell

A PowerShell implementation of FastMCP - An ergonomic interface for the Model Context Protocol (MCP).

Overview

FastMCP for PowerShell provides a streamlined way to work with the Model Context Protocol, enabling efficient interaction with AI models through a clean, PowerShell-native interface. This module helps you create, manage, and utilize model contexts with structured tools, resources, and prompts.

Installation

From PowerShell Gallery (Recommended)

Install-Module -Name FastMCP

Manual Installation

1. Clone this repository
2. Import the module directly:

Import-Module "d:\path\to\FastMCP-PowerShell\FastMCP.psd1"

Quick Start

# Import the module
Import-Module FastMCP

# Create a new server instance and test connectivity
$server = New-FastMCPServer -Endpoint "https://api.example.com/v1" -ApiKey "your-api-key"
if ($server.TestConnection()) {
    Write-Output "Server is reachable!"
} else {
    Write-Output "Server connection failed."
}

# Get a context and add tools, resources, and prompts as needed
$context = Get-FastMCPContext -Server $server

# Simulate sending a request
$response = $context.SendRequest("What is the current weather?")
Write-Output $response.Content

# (Optional) Simulate reporting progress during a long task
for ($i = 10; $i -le 100; $i += 30) {
    $progress = @{ Operation = "Example Task"; PercentComplete = $i; Timestamp = (Get-Date) }
    $context.ReportProgress($progress)
    Start-Sleep -Seconds 1
}

Core Concepts

FastMCP is built around these key concepts:

1. Server - The connection to the underlying AI model API
2. Context - The working environment for interacting with models
3. Tools - Functions that models can use to perform actions
4. Resources - Data assets like images, documents, or databases
5. Prompts - Templates for structuring requests to models

Function Reference

New-FastMCPServer

Creates a new FastMCP server instance that connects to a model provider.

New-FastMCPServer -Endpoint "https://api.openai.com/v1" -ApiKey $env:OPENAI_API_KEY -Provider "OpenAI"

Parameters:

• Endpoint: The API endpoint URL
• ApiKey: Your API authentication key
• Provider: The model provider (optional, defaults to "OpenAI")
• Model: The model to use (optional, defaults to provider's recommended model)
• Options: Additional options as hashtable (optional)

New-Image

Creates an image resource for use with the model.

Example 1: Create an image from a file path

New-Image -Path "./diagram.png" -Description "System architecture diagram"

Example 2: Create a named image (for testing purposes)

New-Image -Name "Diagram" -Description "System architecture diagram"

Example 3: Create an image with both name and path (even if the file doesn't exist)

New-Image -Name "TestImage" -Path "C:\path\to\image.jpg" -Description "Test image" -CreateIfNotExists

Parameters:

• Name: A unique name for the image (used in testing or placeholder scenarios)
• Description: A description of the image content
• Path: File path to the image (used for actual image resources)
• Tags: Optional array of tags for categorization
• CreateIfNotExists: Create an image object even if the file doesn't exist at the specified path

New-Tool

Creates a tool that can be used by the model.

New-Tool -Name "WeatherLookup" -Description "Gets current weather for a location" -Function {
    param($location)
    # Implementation would call a weather API
    return @{ temperature = 72; conditions = "Sunny"; location = $location }
}

Parameters:

• Name: A unique name for the tool
• Description: Description of what the tool does
• Function: ScriptBlock containing the tool's implementation
• Parameters: Optional parameter definitions
• Tags: Optional array of tags

New-Resource

Creates a resource that can be referenced by the model.

New-Resource -Name "UserManual" -Description "Product user manual" -Content $manualText -Type "text" -Tags "documentation"

Parameters:

• Name: A unique name for the resource
• Description: Description of the resource
• Content: The resource content or a path to it
• Type: The resource type (text, json, binary, etc.)
• Tags: Optional array of tags

New-Prompt

Creates a prompt template for structuring requests.

New-Prompt -Name "QueryWithContext" -Description "Basic query with context" -RenderScript {
    param($query, $additionalContext)
    
    return @"
Context information:
$additionalContext

User query: $query

Please provide a helpful response.
"@
}

Parameters:

• Name: A unique name for the prompt
• Description: Description of the prompt's purpose
• RenderScript: ScriptBlock that generates the formatted prompt
• Tags: Optional array of tags

Get-FastMCPContext

Gets a context object for working with the model.

$context = Get-FastMCPContext -Server $server

Parameters:

• Server: The FastMCP server instance to use

Advanced Examples

Creating a Multi-Tool Workflow

# Set up tools for a data analysis workflow
$dataLoader = New-Tool -Name "LoadData" -Description "Loads data from CSV" -Function {
    param($path)
    Import-Csv $path
}

$dataTransformer = New-Tool -Name "TransformData" -Description "Performs data transformations" -Function {
    param($data, $operations)
    # Implementation of data transformations
    # ...
    return $transformedData
}

$dataVisualizer = New-Tool -Name "VisualizeData" -Description "Creates data visualizations" -Function {
    param($data, $type)
    # Implementation of visualization creation
    # ...
    return $visualizationPath
}

# Create the context and add all tools
$context = Get-FastMCPContext -Server $server
$context.AddTool($dataLoader)
$context.AddTool($dataTransformer)
$context.AddTool($dataVisualizer)

# Create an analysis workflow
$response = $context.SendRequest(@"
Please analyze the data in 'sales_data.csv':
1. Load the data
2. Transform it by aggregating by region and quarter
3. Create a bar chart visualization
4. Interpret the results
"@)

Using Resources and Custom Prompts

# Create a resource with company information
$companyInfo = New-Resource -Name "CompanyInfo" -Description "Company background information" -Content @"
Contoso Ltd. is a multinational corporation founded in 2005.
The company specializes in cloud computing solutions and has
over 10,000 employees across 30 countries.
"@ -Type "text"

# Create a prompt template for answering questions with company context
$companyPrompt = New-Prompt -Name "CompanyQuery" -Description "Query with company context" -RenderScript {
    param($query, $companyInfo)
    
    return @"
Company Information:
$companyInfo

Please answer the following question about the company:
$query
"@
}

# Use the context with the resource and prompt
$context = Get-FastMCPContext -Server $server
$context.AddResource($companyInfo)
$context.AddPrompt($companyPrompt)

# Send a request using the prompt
$response = $context.SendRequest("What industry is the company in?", @{
    promptName = "CompanyQuery"
    promptArgs = @{
        companyInfo = $companyInfo.Content
    }
})

Logging Configuration

FastMCP includes a comprehensive logging system:

# Import required type
Add-Type -TypeDefinition @"
public enum LogLevel {
    DEBUG,
    INFO,
    WARNING,
    ERROR,
    CRITICAL
}
"@

# Configure logging
Set-Logging -Level [LogLevel]::DEBUG

# Now all operations will log at DEBUG level
$server = New-FastMCPServer -Endpoint "https://api.example.com/v1" -ApiKey "your-api-key"

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.