assistant

Assitant built with ChatGPT!

Quick start

1. Clone the repo

git clone https://github.com/bjwswang/assistant

2. Install dependencies

cd assistant
go mod tidy

3. Set configuration in assistant.json

see [Configurations](## Configurations) for more details

4. Build the assistant server and CLI

make build

When build is done,you will get two binaries in bin directory.

• assistant is the assistant server
• acli is the assistant CLI which can interact with the assistant server

5. Start the assistant server

./bin/assistant --config assistant.json

6. Test the assistant

curl -XGET http://localhost:9999

Output should be

Welcome to AI Assistant 👋!

Configurations

Parameter	Description	Default
addr	The address which assistant server will watch	:9999
assistant.api_key	OpenAI api key	sk-xxx
assistant.chat.xxx	OpenAI model configuration for Chat	model:gpt-3.5-turbo temperature:0.5 max_tokens:100
assistant.unit_test.xxx	OpenAI model configuration for generating unit tests	model:gpt-3.5-turbo temperature:0.5 max_tokens:100
fiber.xxx	Fiber related parameters	see the official document

APIs

1. Normal chat

• path: /chat
• method: post
• paramters:
  • question: the question you want to chat with the assistant

2. Generate unit test

• path: /ut
• method: post
• paramters:
  • code: the code you want to used for generating unit tests

CLI

1. Chat with assistant

./bin/acli --server http://localhost:9999 chat --question "What is AI assistant in 10 words?"

2. Generate unit tests

./bin/acli --server http://localhost:9999 ut --file {filepath_to_source_code_}"

Contribute to assistant

Welcom to contirbute to this AI assistant!