wickr-bedrock-bot

Wickr Bedrock Bot

This project is a WickrIO integration created via cookiecutter-wickr-bot

Installation

See the WickrIO Getting Started Guide.

Usage

Bedrock Bot understands the following commands:

Available commands in DMs

• /clear - To clear the context in this conversation and start fresh
• /model - To learn what model is currently in use, and switch to a different model in this conversation
• /help - To generate a list of all available commands

Available commands in Rooms

• /chat - To start a conversation with the Bedrock bot in a room, type /chat before the conversation text
• /clear - To clear the context in this chat and start fresh
• /model - To learn what model is currently in use in this conversation, and find out other available models/help - To generate a list of all available commands

Development

Prerequisites

• AWS Account with appropriate permissions to Bedrock. Before you can chat with the bot, you need to enable access to the configured models within your AWS account.
• Additionally, the host machine you're deploying from needs at least 2 CPU cores and 4GB of RAM to successfully run the WickrIOBedrock integration inside the WikcrIO docker container.
• Create an IAM role for your EC2 instance with the following policy:

{"Version": "2012-10-17",

"Statement": [

    {

        "Sid": "VisualEditor0",

        "Effect": "Allow",

        "Action": [

            "bedrock:InvokeModel",

            "bedrock:Retrieve",

            "bedrock:RetrieveAndGenerate",

            "bedrock:GenerateQuery",

            "bedrock:GetKnowledgeBase"

        ],

        "Resource": "*"

    }

]}

• Set up Knowledge base for your bedrock instance

Configuration

Edit model-config-yaml file to include/exclude the models you want to configure for the bedrock bot. Follow the link below to see supported tools https://docs.aws.amazon.com/bedrock/latest/userguide/conversation-inference-supported-models-features.html You can also configure other model parameters.

Edit app-config-yaml file to configure knowledge base id and guidelines accordingly.

Building

Run make to create a new software.tar.gz package, which can be installed as a WickrIO Custom Integration. See the WickrIO docs for more details.