Analysis updated 2026-05-18
Search a codebase for concepts like error handling without needing exact keyword matches.
Let an AI coding agent such as Claude Code search your repository with semantic understanding.
Run NextPlaid as a standalone REST API to power semantic search in another project.
| lightonai/next-plaid | helvesec/rmux | gi-dellav/zerostack | |
|---|---|---|---|
| Stars | 422 | 391 | 459 |
| Language | Rust | Rust | Rust |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 3/5 | 4/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Runs as a local binary with no external server, but the underlying NextPlaid API deployment uses Docker.
This repository contains two related tools: ColGREP and NextPlaid. ColGREP is a command-line search tool for codebases that combines two search techniques, regular expression pattern matching and semantic meaning-based search. Instead of just finding lines that match a text pattern, it understands what you are looking for conceptually, so you can search for "error handling" and find relevant functions even if those exact words don't appear in the code. Everything runs locally as a single binary with no server or external service required, your code never leaves your machine. ColGREP integrates with AI coding agents like Claude Code, OpenCode, and Codex, allowing those tools to search your codebase using semantic understanding. It works by parsing your code with Tree-sitter (a code analysis tool), creating structured descriptions of each function and class, and then building a multi-vector index, a search index that stores many small representations of each code unit rather than just one, making matches more accurate for complex queries. NextPlaid is the underlying search database that powers ColGREP. It is a general-purpose multi-vector search engine available as a standalone REST API (accessible via Docker), with a Python client for integration into other projects. It supports compressed storage formats to keep memory usage manageable, incremental updates without rebuilding the full index, and SQL-style filtering to narrow results before the semantic search runs. Both tools are designed to run efficiently on CPU hardware.
A local, no-server code search tool combining regex and semantic search, built on a multi-vector search database called NextPlaid.
Mainly Rust. The stack also includes Rust, Tree-sitter, Docker.
No license information is stated in the explanation.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.