LLM fixer
A tool that automatically fixes static analysis errors.
Installation
Add this line to your Gemfile:
gem 'llm_fixer'And then execute:
$ bundle installOr install it yourself as:
$ gem install llm_fixerEnvironment Variables
You can customize the behavior using the following environment variables:
-
LLM_API_KEY: LLM API key -
LLM_API_BASE: Base URL for the LLM API -
LLM_MODEL: LLM model to use
Usage
Set the environment variables:
ChatGPT
export LLM_API_KEY="your-api-key-here"
export LLM_API_BASE="https://api.openai.com/v1"
export LLM_MODEL="gpt-4o"Google Gemini
export LLM_API_KEY="your-api-key-here"
export LLM_API_BASE="https://generativelanguage.googleapis.com/v1beta"
export LLM_MODEL="gemini-2.0-pro-exp"Local(LM Studio)
export LLM_API_KEY="n/a"
export LLM_API_BASE="http://localhost:1234"
export LLM_MODEL="tinyswallow-1.5b-instruct"Run your static analysis tool command (e.g., RuboCop):
llm_fixer fix your-lint-or-test-command path/to/target
# RuboCop example
llm_fixer fix bundle exec rubocop path/to/target.rb
# With additional prompt to guide the LLM
llm_fixer fix --additional-prompt "Follow the style guide and use double quotes" bundle exec rubocop path/to/target.rb
# or using the short form
llm_fixer fix -a "Follow the style guide and use double quotes" bundle exec rubocop path/to/target.rbThe --additional-prompt (or -a) option allows you to provide additional instructions to the LLM when fixing errors. This can be useful when you want to customize the style or approach used for fixes.