/infraGen

AI Assisted System Architecture Design

Primary LanguageJupyter NotebookMIT LicenseMIT

Infragen

Your AI sparring partner for cloud infrastructure

Screenshot 2024-02-12 at 16 00 17

Installing

The app is not ready for production usage or direct installation form the VSCode extension store. For now, if you want to use InfraGen, you have to set up the development environment.

Requirements:

  • Python 3.11
  • Conda
  • Node 20
  • pnpm (npm i -g pnpm)
  • VSCode
  • Inframap (link)
  • Graphviz CLI (link)

  1. Install JavaScript dependencies: run pnpm i and (cd webview-ui && pnpm i && pnpm build)

  2. Create a .env file in the root folder by copying and renaming the .env.example file and filling in the required values

  3. Install python dependencies, run conda env create -f environment.yml in the ./server folder

  4. Add your own example terraform files to the ./server/examples folder

  5. To generate the embedding database, run python generate_embedding_db.py in the ./server folder

  6. (temporary) There is, at the time of writing, a bug in Langchain which prevents the creation of a custom message template in the server/lc_agent.py file. To fix this, you have to manually edit the code where the error occurs. Either wait for the error to occur when running the application for the first time, or edit the /Users/{your_username}/miniconda3/envs/infragen/lib/python3.11/site-packages/langchain/agents/openai_tools/base.py file and remove the following lines 81-83:

      missing_vars = {"agent_scratchpad"}.difference(prompt.input_variables)
  if missing_vars:
      raise ValueError(f"Prompt missing required variables: {missing_vars}")

Running

  1. Start the redis instance: run docker compose up in the ./server folder
  2. Start the backend server: run uvicorn main:app --reload in the ./server folder
  3. Run the extension: Open this folder in VSCode, press f5 to launch the task to start a new VSCode window with the plugin installed
  4. Open an extension window by opening the command menu (⌘ + Shift + P) and selecting the Run Infragen command

Architecture

architecture

The extension requires a backend, which runs locally for now. The backend takes care of running the LLM-Agent and accessing all requried providers like the OpenAI API, a local version of ChromaDB and a Redis instance for persisting chat history and modified code.