The most recent message in the conversation was not being used as context.

#127

Implemented asynchronous runtime to improve performance, and also made other performance improvements to the mf diff functionality which also improves performance and quality on the commit and pr commands. This release also comes with a big refactoring of the code base, and is starting to separate out the client-side logic (CLI) from the core library - i.e. the stuff for document segmentation, chat history management, indexing, and managing API requests to GPT.

• Small refactors
• Improved ConfiguredModel and MindFlowModel abstractions for clarity, extensibility, and performance.

Big Release! This one deserves 0.4.0.

• Improved indexing and query performance.
• Use Pinecone to store vectors for speed and scale.
• Refactor the codebase to be more clean and generic.
• Add support for Anthropic's Claude and Claude Instance models.

• Added GPT 4 and configuration for the model used to generate MindFlow responses.
• Improved performance and generalization of the object model.