Which AI Agent or Code Editor Best Understands Code Context? (Comparison for Small, Mid-size, and Large Codebases)

I haven't found an LLM that can analyze the entire codebase across multiple files and understand the context that well. I'm using Cursor with Claude 3.7 right now, and the codebases I've been working on are primarily mid-size React projects.

Here's what has been working really well for me:

• Autocompleting and updating references
• Creating utility and helper functions with a clearly defined input and output
• Stubbing out new components
• Writing out comments and having the LLM come up with functions
• Learning a new code base by highlighting confusing logic and asking what the code does
  • (this one works really well for me because Clause will show example inputs and outputs)

But, from the above points, you'll see that the LLM mainly works well within a single file.

Windsurf is the AI IDE for better large context management. It's much better than cursor, but I havent messed around with it too much. I'm still sticking to cursor because I find that the QOL improvement from the code completion to be much higher ROI than asking LLMs to write code.

The code completion is usually doing boilerplate tasks that allow me to focus on the logic instead of "what did i name this variable" or "which file did function X belong to again". Or if I convert a list to a set, then it automatically updates the syntax in existing code to the set syntax instead of manually doing it

See this thread on why windsurf is better from the founder: x.com/_mohansolo/status/1899630153636118529

The TL;DR is they spent a ton of effort and lot more compute per query to find the right files for your query