文件列表:

Babulus-production/ (0, 2023-08-15)
Babulus-production/.bundle/ (0, 2023-08-15)
Babulus-production/.bundle/config (48, 2023-08-15)
Babulus-production/Guardfile (78, 2023-08-15)
Babulus-production/Rakefile (267, 2023-08-15)
Babulus-production/app/ (0, 2023-08-15)
Babulus-production/app/Gemfile (282, 2023-08-15)
Babulus-production/app/default_openai_system_prompt.txt (3686, 2023-08-15)
Babulus-production/app/functions/ (0, 2023-08-15)
Babulus-production/app/functions/check_bot_health.rb (12348, 2023-08-15)
Babulus-production/app/functions/get_bot_setting.rb (1098, 2023-08-15)
Babulus-production/app/functions/get_bot_setting.spec (961, 2023-08-15)
Babulus-production/app/functions/get_system_information.rb (670, 2023-08-15)
Babulus-production/app/functions/get_system_sales_performance_information.rb (1005, 2023-08-15)
Babulus-production/app/functions/set_bot_setting.rb (1176, 2023-08-15)
Babulus-production/app/functions/set_bot_setting.spec (527, 2023-08-15)
Babulus-production/app/handle_alarm_notifications.rb (2895, 2023-08-15)
Babulus-production/app/handle_alarm_notifications.spec (0, 2023-08-15)
Babulus-production/app/handle_api_request.rb (1915, 2023-08-15)
Babulus-production/app/handle_api_request.spec (3119, 2023-08-15)
Babulus-production/app/lib/ (0, 2023-08-15)
Babulus-production/app/lib/babulus.rb (557, 2023-08-15)
Babulus-production/app/lib/babulus/ (0, 2023-08-15)
Babulus-production/app/lib/babulus/cloudwatch_metrics.rb (3280, 2023-08-15)
Babulus-production/app/lib/babulus/configuration/ (0, 2023-08-15)
Babulus-production/app/lib/babulus/configuration/setting.rb (2677, 2023-08-15)
Babulus-production/app/lib/babulus/configuration/setting.spec (1840, 2023-08-15)
Babulus-production/app/lib/babulus/function.rb (1363, 2023-08-15)
Babulus-production/app/lib/babulus/function.spec (528, 2023-08-15)
Babulus-production/app/lib/babulus/key_value_store.rb (1555, 2023-08-15)
Babulus-production/app/lib/babulus/key_value_store.spec (2195, 2023-08-15)
Babulus-production/app/lib/babulus/logger_factory.rb (252, 2023-08-15)
Babulus-production/app/lib/babulus/logger_factory.spec (1024, 2023-08-15)
Babulus-production/app/lib/babulus/openai_chat_bot.rb (14307, 2023-08-15)
Babulus-production/app/lib/babulus/openai_chat_bot.spec (99078, 2023-08-15)
... ...

# Babulus A babbling bot that can do things for you. Uses the Open AI chat API with function calling. Currently available through Slack, and soon through Microsoft Teams. ## An AI agent framework Babulus is a framework for building serverless AI agents in the form of infrastructure-as-code applications managed by AWS CloudFormation. Each deployed bot includes a key-value store in DynamoDB for its settings caching things and a message cache table for caching the conversation history. Each bot instance also includes a CloudWatch dashboard showing high-level application metrics and it can show you metric widget images from that dashboard if you ask in the chat. These features are demonstrations of how to use the function plugin system to give the agent access to your own app's CloudWatch metrics and alerts so that you can set up a serverless copilot for your system. What good are metrics and alerts without notifications? And so, a Babulus bot's alerts are connected to an SNS topic that will trigger a Lambda function that will post a notification about the alert state changing to a Slack channel that you specify. You can connect your own alarms to that SNS topic to get notified about them, too. After you set up a Slack app and give it your bot's Slack Events API handler URL, messages from Slack will trigger your bot's Lambda functions through API Gateway and through an SQS queue for resliliency. Your bot will then sit between Slack and the OpenAI API and will use its current settings to craft requests and send them to the chat API. The art in a chat bot is in how you manage the tokens that you send to the language model, so the "crafting" part is important. You can start by customizing your default system prompt, which is a parameter in AWS Systems Manager. ## Calling chat functions As a demonstration of function calling, a Babulus bot can tell you about its own CloudWatch metrics and show you metric widget images. It has a plugin architecture for adding custom chat functions for doing whatever you need. ## Overview The application is based on AWS Lambda and is triggered by events from the Slack Events API. This means that whenever a user posts a message in the linked Slack channel, an event is sent to our Lambda function, which processes the event and sends a response back to the Slack channel. The core processing logic is encapsulated in a Ruby class called `Babulus::SlackEventsAPIHandler`. ```mermaid graph TD A(Slack User) -- 1: sends message --> B((Slack API)) B -- 2: triggers message event --> C[AWS Lambda\nin API Gateway] C -- 3: posts message --> F((SQS Queue)) C -- 4: responds successfully\nimmediately --> B F -- 5: triggers event --> G[AWS Lambda\non SQS event] G -- 6: executes --> D[SlackEventsAPIHandler] D -- 7: sends slack message history --> E[OpenAIChatBot] D -- 8: gets channel message history --> B E -- 9: sends chat request with channel history --> H((OpenAI API)) H -- 10: sends chat response --> E E -- 11: posts response to Slack channel --> B B -- 12: posts message --> A ``` The `Babulus::SlackEventsAPIHandler` class is responsible for parsing incoming Slack events, dispatching them to the correct handler based on their type, and returning the appropriate response. Currently, it supports two types of events: `url_verification` and `message`. The `url_verification` event is used by Slack to confirm that our endpoint is operational and owned by us. When this event type is received, the class simply echoes back the 'challenge' string provided in the event. The `message` event type is triggered whenever a message is posted in the Slack channel. When this event type is received, the class logs the message text and sends a response back to the Slack channel. Currently, the response is simply a mirror of the original message, but the plan is to replace this with a call to the OpenAI API, so the response will be generated by GPT-4. ## Configuration The application is configured to run in different environments (e.g., development, production) using configuration parameters stored in AWS Systems Manager (SSM) Parameter Store, segmented by environment name. Your Slack app access token and your OpenAI API token are secret, and so those go in Secrets Manager. ## Operation ### Deployment Replace the profile with your correct AWS profile, and provide the correct stack name for your environment. #### Production $ sam build && sam deploy --profile=330731243300_AdministratorAccess --region us-east-1 --config-env=production #### Development $ sam sync --profile=AdministratorAccess-330731243300 --region us-east-1 --stack-name=babulus-development ### View logs $ sam logs -t --profile=AdministratorAccess-330731243300 --region=us-east-1 --stack-name=babulus-development ## Development ### Local invocation $ sam build && sam local invoke SlackEventsAPIHandlerHandlerFunction --event=events/slack_events_api_test_message.json --env-vars=env.json --profile=AdministratorAccess-330731243300 ### Testing The code includes a comprehensive suite of RSpec tests to ensure that it handles Slack events correctly. These tests cover the main event dispatching logic, as well as the individual handlers for the supported event types. bundle exec rake spec You can use Guard to watch the code and run the specs when anything changes: bundle exec guard ## Contribution Contributions to this project are welcome! Feel free to fork the repository and submit a pull request with your changes. Before submitting your pull request, please ensure that your changes pass the test suite and that you have added tests for any new functionality. Also, please update the README if necessary to reflect your changes. ## Credits Most of the code and the README were written by ChatGPT, an AI model developed by OpenAI. You can view the conversation that resulted in this README [here](https://chat.openai.com/share/294e63c7-3398-4dc6-a353-6435c23fcd6a).

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：