pwshBedrock

What is pwshBedrock?

pwshBedrock is a PowerShell module designed to facilitate interaction with foundation models (FMs) on the Amazon Bedrock platform. It provides a comprehensive set of cmdlets that allow users to send messages, retrieve responses, manage conversation contexts, generate and transform images, and estimate costs. With support for both direct model calls via InvokeModel and a consistent interface through the Converse API, pwshBedrock simplifies the integration of generative AI capabilities into PowerShell scripts and workflows, making it accessible and efficient for developers to leverage AI in their PowerShell projects.

Why pwshBedrock?

Interacting with foundation models on Amazon Bedrock typically requires extensive research to understand each model's unique capabilities and the precise JSON payloads needed for interaction. This can be time-consuming and complex, especially when crafting payloads and converting data formats isn't the primary focus for many PowerShell developers.

pwshBedrock abstracts this complexity by supporting parameters for each model based on their capabilities, allowing you to pass simple values and native PowerShell objects. The module handles the JSON payload formation, enabling you to interact with models using straightforward commands without delving into the intricacies of payload structures.

• Converse API Made Easy: For Converse API interactions, pwshBedrock eliminates the need to manually create and manage complex Amazon Bedrock Runtime objects. Standard parameters and PowerShell objects can be used, with all necessary conversions handled internally by the module.
• Context Management: Unlike native interactions, pwshBedrock natively manages ongoing context with models, storing interaction histories in memory. It also supports saving and retrieving conversation contexts from disk, allowing you to maintain continuous and coherent interactions across sessions.
• Media and File Handling: The module simplifies media and document interactions by handling base64 encoding and memory stream conversions as required by the models. It also supports specifying save paths for generated images, managing retrieval and conversion seamlessly.
• Token Tracking and Cost Estimation: pwshBedrock tracks token usage and provides basic cost estimation, giving you insights into your model interactions' usage and cost implications.
• Model Capability Insights: The module stores model support information, allowing you to quickly retrieve a model's capabilities and understand what each model can do without extensive research.

Getting Started

Prerequisites

• PowerShell 5.1 or later
• AWS account with access to Amazon Bedrock
  • AWS credentials with appropriate bedrock:InvokeModel permission
  • You must manage and add model access through one-time request model access action. Note: this must be done for each model you plan to interact with.

Installation

# Install pwshBedrock from the PowerShell Gallery
Install-Module -Name pwshBedrock -Repository PSGallery -Scope CurrentUser

Quick start

Converse

Converse provides a consistent interface that works with most models that support messages.

#------------------------------------------------------------------------------------------------
# Sends a message to the specified model via the Converse API in the specified AWS region and returns the response.
Invoke-ConverseAPI -ModelID anthropic.claude-3-5-sonnet-20240620-v1:0 -Message 'Explain zero-point energy.' -Credential $awsCredential -Region us-east-1
#------------------------------------------------------------------------------------------------

Models

AI21 Labs Jamba-Instruct models

#------------------------------------------------------------------------------------------------
# Sends a chat message to the on-demand AI21 Labs Jamba model in the specified AWS region and returns the response.
Invoke-AI21LabsJambaModel -Message 'Explain zero-point energy.' -ModelID 'ai21.jamba-1-5-large-v1:0' -Credential $awsCredential -Region 'us-east-1'
#------------------------------------------------------------------------------------------------

Amazon Image Generator

#------------------------------------------------------------------------------------------------
# Generates an image and saves the image to the C:\temp folder.
Invoke-AmazonImageModel -ImagesSavePath 'C:\temp' -ImagePrompt 'Create a starship emerging from a nebula.' -ModelID 'amazon.nova-canvas-v1:0' -Credential $awsCredential -Region 'us-east-1'
#------------------------------------------------------------------------------------------------

Amazon Titan Text models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Amazon Titan model in the specified AWS region and returns the response.
Invoke-AmazonTextModel -Message 'Explain zero-point energy.' -ModelID amazon.titan-text-lite-v1 -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Amazon Nova Reel models

#------------------------------------------------------------------------------------------------
# Sends messages to an Amazon Nova Reel model on the Amazon Bedrock platform to generate a video.
Invoke-AmazonVideoModel -VideoPrompt 'Closeup of a large seashell in the sand, gentle waves flow around the shell. Camera zoom in.' -ModelID 'amazon.nova-reel-v1:1' -S3OutputURI 's3://mybucket' -Credential $awsCredential -Region 'us-east-1'
#------------------------------------------------------------------------------------------------
# Generates a video asynchronously using the Nova Reel AI model and stores the output in the specified S3 bucket. Downloads the video to the specified local path.
Invoke-AmazonVideoModel -VideoPrompt 'A cat playing with a ball.' -ModelID 'amazon.nova-reel-v1:1' -S3OutputURI 's3://mybucket' -AttemptS3Download -LocalSavePath 'C:\temp\videos' -Credential $awsCredential -Region 'us-east-1'
#------------------------------------------------------------------------------------------------

Amazon Nova Text models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Amazon Nova model in the specified AWS region and returns the response.
Invoke-AmazonNovaTextModel -Message 'Explain zero-point energy.' -ModelID 'amazon.nova-pro-v1:0' -Credential $awsCredential -Region 'us-east-1'
#------------------------------------------------------------------------------------------------

Anthropic Models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Anthropic model in the specified AWS region and returns the response.
Invoke-AnthropicModel -Message 'Explain zero-point energy.' -ModelID 'anthropic.claude-sonnet-4-20250514-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------
# Sending a media file to an Anthropic model and retrieving the response
Invoke-AnthropicModel -Message 'What can you tell me about this picture?' -ModelID 'anthropic.claude-sonnet-4-20250514-v1:0' -MediaPath 'C:\images\tanagra.jpg' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Cohere Command models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Cohere Command model in the specified AWS region and returns the response.
Invoke-CohereCommandModel -Message 'Explain zero-point energy.' -ModelID 'cohere.command-text-v14' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Cohere Command R and Command R+ models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Cohere Command R model in the specified AWS region and returns the response.
Invoke-CohereCommandRModel -Message 'Explain zero-point energy.' -ModelID 'cohere.command-r-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

DeepSeek Models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Cohere Command R model in the specified AWS region and returns the response.
Invoke-DeepSeekModel -Message 'Explain zero-point energy.' -ModelID 'deepseek.r1-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Luma AI Models

#------------------------------------------------------------------------------------------------
# Sends messages to a Luma AI model on the Amazon Bedrock platform to generate a video.
Invoke-LumaAIModel -VideoPrompt 'A cat playing with a ball' -S3OutputURI 's3://mybucket' -ModelID 'luma.ray-v2:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------
# Generates a video asynchronously using the Luma AI model and stores the output in the specified S3 bucket. Downloads the video to the specified local path.
Invoke-LumaAIModel -VideoPrompt 'A cat playing with a ball' -S3OutputURI 's3://mybucket' -ModelID 'luma.ray-v2:0' -AttemptS3Download -LocalSavePath 'C:\temp\videos' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Meta Llama models

#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Meta model in the specified AWS region and returns the response.
Invoke-MetaModel -Message 'Explain zero-point energy.' -ModelID 'meta.llama4-scout-17b-instruct-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------
# Sending a media file to a Meta model and retrieving the response
Invoke-MetaModel -ImagePrompt 'Describe this image in two sentences.' -ModelID 'meta.llama4-scout-17b-instruct-v1:0' -MediaPath 'C:\path\to\image.jpg' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Mistral AI models

#------------------------------------------------------------------------------------------------
# Sends a chat message to the on-demand Mistral AI chat model in the specified AWS region and returns the response.
Invoke-MistralAIChatModel -Message 'Explain zero-point energy.' -ModelID 'mistral.mistral-large-2402-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------
# Sends a text message to the on-demand Mistral AI text model in the specified AWS region and returns the response.
Invoke-MistralAIModel -Message 'Explain zero-point energy.' -ModelID 'mistral.mistral-large-2402-v1:0' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------

Stability AI models

#------------------------------------------------------------------------------------------------
# Generates an image using the on-demand StabilityAI diffusion image model. The returned image is saved to the provided output path.
Invoke-StabilityAIDiffusionXLModel -ImagesSavePath 'C:\temp' -ImagePrompt 'Create a starship emerging from a nebula.' -ModelID 'stability.stable-diffusion-xl-v1' -Credential $awsCredential -Region 'us-west-2'
#------------------------------------------------------------------------------------------------
# Generates an image using the on-demand StabilityAI image model. The returned image is saved to the provided output path.
Invoke-StabilityAIImageModel -ImagesSavePath 'C:\images' -ImagePrompt 'Create a starship emerging from a nebula.' -ModelID 'stability.stable-image-ultra-v1:0' -Credential $awsCredential -Region 'us-west-2'
#-----------------------------------------------------------------------------------------------

General

#------------------------------------------------------------------------------------------------
# Returns the message context history for the specified model.
Get-ModelContext -ModelID 'anthropic.claude-3-5-sonnet-20241022-v2:0'
#------------------------------------------------------------------------------------------------
# Estimates the cost of using the model with 1000 input tokens and 1000 output tokens.
Get-ModelCostEstimate -InputTokenCount 1000 -OutputTokenCount 1000 -ModelID 'cohere.command-r-plus-v1:0'
#------------------------------------------------------------------------------------------------
# Gets information about specified model
Get-ModelInfo -ModelID 'anthropic.claude-3-opus-20240229-v1:0'
# Retrieves information for all models.
Get-ModelInfo -AllModels
#------------------------------------------------------------------------------------------------
# Retrieves the tally for the model
Get-ModelTally -ModelID 'meta.llama3-70b-instruct-v1:0'
#------------------------------------------------------------------------------------------------
# Estimates the number of tokens in the provided text.
Get-TokenCountEstimate -Text 'Explain zero-point energy.'
#------------------------------------------------------------------------------------------------
# Resets the message context for the specified model.
Reset-ModelContext -ModelID 'anthropic.claude-v2:1'
#------------------------------------------------------------------------------------------------
# Resets the tally for the model
Reset-ModelTally -ModelID 'mistral.mistral-large-2402-v1:0'
#------------------------------------------------------------------------------------------------
# Saves the message context history for the specified model to a file,
Save-ModelContext -ModelID 'amazon.titan-tg1-large' -FilePath 'C:\temp'
#------------------------------------------------------------------------------------------------
# Loads and sets the message context for a model from a file.
Set-ModelContextFromFile -FilePath 'C:\temp\context.xml'
#------------------------------------------------------------------------------------------------

How pwshBedrock Works

pwshBedrock leverages the AWS.Tools.BedrockRuntime from the AWS Tools for PowerShell to interact with on-demand foundational models in your AWS account.

• Core Functionality:
  • InvokeModel Calls: Uses the Invoke-BDRRModel cmdlet to send commands to the models.
  • Converse API Calls: Uses the Invoke-BDRRConverse cmdlet to manage interactions via the Converse API.
• Simplified Parameter Handling: pwshBedrock simplifies the process by providing easy-to-use parameters that automatically form the required payloads for both Invoke-BDRRModel and Invoke-BDRRConverse. This includes built-in validations and checks to ensure that only supported values are passed.
• Context Management - For models that support conversation context, pwshBedrock automatically manages this context:
  • In-Memory Storage: The context of the model interaction is stored in memory by default.
  • File System Storage: You can also save the context to the file system and retrieve it later, ensuring continuous interactions across sessions.

Features

Note: Not all models support all features

• Messaging: Directly engage with Bedrock models using simple or custom messages and receive responses.
• Custom Conversations: Supports custom formats for conversations, allowing for tailored interactions.
• Context Control: Maintain continuous and coherent interactions with models through effective context management.
• Image Generation: Leverage powerful image generative capabilities to both create and transform images.
• Video Generation: Leverage powerful video generative capabilities to create videos.
• Document Support: Upload common document types directly to the model to provide context or generate summaries.
• Vision:: Provide medial files directly to the model for context and evaluation.
• Cost Estimation: Estimates the cost of your AI engagements, helping you keep a tab on model engagement.
• Function calling: Connect models to external tools. Enables models to handle specific use cases and pull custom responses.
• Converse API support Provides a consistent interface that works with most models that support messages. This allows you to write code once and use it with different models. It also provides a consistent response format for each model.