VectorCode

Why we included this project

Language models give their worst answers on code they never saw during training: closed-source repositories, projects too young to have much of a footprint, and libraries still in active development. VectorCode is an indexing tool built to fill that hole by building a searchable index of the repository you're working in and letting you pull the task-relevant files and context into the prompt before you call the model. The same index is reachable from a command-line tool, an MCP server, and a Neovim plugin, so the retrieval can feed a standalone script or an editor plugin you already use, with a Lua API for building custom Neovim workflows. That makes it a natural fit for developers on proprietary or fast-moving codebases who would rather hand the model accurate context than accept a guessed answer, and the project is upfront about being beta and changing quickly, so budget for interface churn while you evaluate it.