This is really interesting because I am working on pretty large codebases simultaneously and the idea that my agent won't have to re-index and re-read all my files over and over again is pretty enticing.
My question is what methods are you using to index and connect these to your agents?
My research has led me to the following but I am basically very confused by what metric to measure how good they are.
1. https://github.com/abhigyanpatwari/GitNexus - seems like the most popular
2. https://github.com/DeusData/codebase-memory-mcp - also interesting/new
3. https://github.com/JaredStewart/coderlm - the tree-sitter based approach seems really good.
Would love to know how people actively using frontier models and methods navigate this domain.
Serena gives you symbol-level navigation, reading, and editing via tree-sitter. So you're already getting much of what CoderLM offers — plus Serena has editing capabilities(replace symbol body, insert before/after symbol, rename). GitNexus goes further with the graph/impact analysis layer on top.
I personally use Serena and GitNexus.
The case for using both
- GitNexus tells you what to change (impact analysis, dependency maps, blast radius)
- Serena makes the change (symbol-level editing, safe renames)
That's a strong workflow: understand first, then act precisely.