RAG Frameworks · Davidyz

VectorCode

VectorCode is a Python-based code repository indexing tool that builds vector embeddings of your codebase to improve LLM prompts and reduce hallucination. It provides a CLI, Neovim plugin, and MCP server integration for retrieval-augmented generation (RAG) workflows with code context.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/Davidyz/VectorCode

871

GitHub stars

Forks

Python

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	Davidyz/VectorCode
Owner	Davidyz
Primary language	Python
License	MIT — OSI-approved
Stars	871
Forks	49
Open issues	18
Latest release	0.7.20 (2025-12-15)
Last updated	2026-02-23
Source	https://github.com/Davidyz/VectorCode

What VectorCode is

VectorCode indexes code repositories using embeddings and semantic chunking (via tree-sitter), stores vectors in persistent Chroma databases, and exposes query APIs through CLI, Lua/Neovim plugin, and MCP server interfaces. It respects .gitignore, auto-detects project roots via .git/.vectorcode.json, and supports configurable chunk sizes and file filtering.

Quickstart

Get the VectorCode source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/Davidyz/VectorCode.git2cd VectorCode3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Proprietary or Closed-Source Codebase RAG

Inject task-relevant context from internal projects into LLM prompts where training data is incomplete or absent, significantly reducing model hallucination on unfamiliar codebases.

Neovim-Based AI-Assisted Development Workflow

Integrate with Neovim plugins (codecompanion.nvim, minuet-ai.nvim) via MCP or Lua API to augment code generation, refactoring, and documentation tasks with live repository context.

Multi-Language Repository Context Retrieval

Support semantic chunking across multiple languages (via tree-sitter) for polyglot projects, enabling precise code fragment retrieval for prompt construction.

Implementation considerations

Plan for initial indexing time and Chroma storage overhead; tree-sitter chunking adds CPU cost but improves retrieval quality. Monitor `.vectorcode/` directory size in large repos.
Verify tree-sitter language parser availability for all languages in your codebase; fallback chunking behavior for unsupported languages is not clearly documented.
Test `.vectorcode.json` project-root detection and `.gitignore` handling to ensure correct file inclusion/exclusion before production use.
Check integration compatibility with your target Neovim plugin ecosystem (codecompanion.nvim, minuet-ai.nvim) and MCP version support.
Set version constraints (e.g., `vectorcode<0.7.0`) if breaking changes in minor versions are unacceptable; plan upgrade cycles aligned with release notes.

When to avoid it — and what to weigh

Requires Real-Time or Production Model Serving — VectorCode is a local indexing and retrieval tool; it does not fine-tune or serve models. Use only for augmenting LLM inputs, not for replacing model inference pipelines.
Enterprise Authentication and Access Control Required — Remote Chroma DB support without proper authentication is marked as TODO. Centralized, multi-user deployments with fine-grained access control are not yet supported.
Very Large Repositories with Strict Latency SLAs — Initial indexing and chunking overhead may be significant for massive codebases. Retrieval performance at scale and optimization for sub-second queries are not documented.
Stable, Production-Grade Interfaces Needed — Project is explicitly in beta; minor version bumps introduce breaking changes. Not recommended for teams with strict API stability requirements until 1.0.0 release.

License & commercial use

Licensed under MIT (MIT License), a permissive open-source license that allows unrestricted commercial use, modification, and distribution with attribution required.

MIT license permits commercial use without restriction. However, as a library/tool, commercial viability depends on your use case (internal tooling vs. product offering). No warranty or support guarantees from the project; review project maturity (beta status) before relying on it in revenue-critical workflows.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Good
Assessment confidence	High

Security considerations

Runs locally with access to your codebase files. Chroma persistent-client stores vectors on disk; ensure file permissions restrict access. Remote Chroma lacks authentication (TODO), so do not expose to untrusted networks. No explicit vulnerability disclosure process mentioned; review GitHub for reported security issues.

Alternatives to consider

Codebase Retrieval (e.g., Sourcegraph, CodeSearch)

Sourcegraph and CodeSearch offer scalable, hosted code indexing but are not LLM-focused RAG tools and do not natively integrate embedding vectors for prompt augmentation.

LLamaIndex / LangChain Document Loaders

These frameworks provide generic code chunking and embedding, but require more manual setup and offer less Neovim/MCP integration than VectorCode's opinionated workflow.

Copilot, Cursor, or Built-In IDE AI Features

Commercial IDEs and editors offer native RAG for context. VectorCode is open-source and extensible but requires manual setup; trade-off is control vs. ease.

Software development agency

Build on VectorCode with DEV.co software developers

Start with VectorCode CLI to index your repository, then integrate with Neovim or MCP for AI-assisted development. Review the documentation and test with a non-critical project first.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

embedJs code-graph-rag RAGLight

Related on DEV.co

Explore the category and the services that help you build with it.

RAG Frameworks AI Development Custom Software Talk to DEV.co

VectorCode FAQ

Does VectorCode fine-tune or train models?

No. VectorCode only indexes and retrieves code snippets. You supply the LLM (via your own API key or integration); VectorCode augments the prompt with relevant context.

Can I use VectorCode with VS Code or other editors?

Currently, Neovim integration is the primary editor. MCP server support exists; VS Code users can integrate via MCP clients if available. Other editor adapters are not documented.

How do I set up remote Chroma with VectorCode?

Remote Chroma support without authentication is marked as TODO. Currently, only local persistent-client is production-ready. Manual remote setup may require workarounds.

What happens if a language is not supported by tree-sitter?

Fallback chunking behavior is not explicitly documented. Review the codebase or file an issue to confirm handling for unsupported languages before indexing critical code.

Software development & web development with DEV.co

DEV.co helps companies turn open-source tools like VectorCode into production software. Our software development services cover the full lifecycle — architecture, web development, integration, and maintenance — delivered by software developers and web developers who ship. Engage our software development agency to implement or customize it for your rag frameworks stack.

Ready to Index Your Codebase?

Start with VectorCode CLI to index your repository, then integrate with Neovim or MCP for AI-assisted development. Review the documentation and test with a non-critical project first.

Talk to DEV.co Schedule a Consultation