configurable default model

[pelikhan]

Environment variable to specify the default model for all scripts

• 🔄 The default model for the GenAIScript has been changed from "gpt-4" to "openai:gpt-4".
• 🎛️ Introduced the ability to set the default model and temperature via environment variables (GENAISCRIPT_DEFAULT_MODEL and GENAISCRIPT_DEFAULT_TEMPERATURE).
• 🎚️ The default temperature has been changed from 0.2 to 0.8.
• 🐛 Multiple assertions have been added to ensure that the modelId is defined before it is used.
• 📚 Documentation updates have been made to reflect these changes and provide additional context and usage examples.
• 🧰 The NodeHost.install() function has been updated to be async, and now includes a call to parse default settings from the environmental variables.
• ✔️ Extra validation has been added in the githubCreatePullRequestReview function to check the distance between lines when comparing with existing comments.
• ⚙️ Several functions in the expandTemplate module have been updated to use the new default settings from the host.
• ✨ The parseTokenFromEnv function now handles parsing the default model from the environment variables.
• 🔧 The .env configuration handling has been improved, with additional checks and balances.

generated by pr-describe

[pelikhan]

/genai-describe

[pelikhan]

/genai-review

[github-actions]

• The code changes seem to primarily focus on improving model configuration and handling in the application, with specific updates to handle default model configurations.
• The code now handles configuration defaults for the model and temperature from environment variables, which increases the flexibility of the software.
• The code also introduces changes to error handling for model selection. Now, if a model is not specified, the code asserts for it and falls back to the default model specified in the host.
• The code includes improvements in the generation of the .env file template.
• There are also updates to test configurations, and several modifications to adapt to the model changes.

However, there are a couple of areas which could be improved:

1. In the cli.ts file, the async keyword is added to the preAction hook. However, it's unclear if the parent function that calls this hook can handle promises or async operations. If the parent function can't handle promises, this could lead to unhandled promise rejections.

2. In the constants.ts file, the GITHUB_PULLREQUEST_REVIEW_COMMENT_LINE_DISTANCE constant is added but it's not clear how this constant is used. If this is an important configuration value, it might be better to put it into a configuration file or in environment variables.

Otherwise, LGTM 🚀.

generated by pr-review