ez-search
Semantic search. Zero cloud.
webgpu accelerated · 100% local · open source
Search by meaning
ez-search is a semantic codebase search CLI. Search by meaning, not strings. Embeddings run on your GPU (CPU fallback included), so your code never touches a network. Built as a retrieval engine for AI coding assistants like Claude Code.
What's inside
Six things, no fluff. Everything runs on your machine.
100% Local
Embedding and search run on your machine. No API keys. No network calls. Your code never leaves your device.
Three Pipelines
Dedicated models for code, text/docs, and images. Search the whole codebase by meaning.
GPU-Accelerated
WebGPU first, CPU fallback automatic. GPU just makes it faster.
Incremental Indexing
Only re-embeds files that changed. Re-indexing is instant for unchanged files.
Built for AI
JSON output shaped for AI consumption. Drops into Claude Code as a retrieval tool.
Zero Config
Works out of the box. Respects .gitignore, skips node_modules and lockfiles automatically.
Questions
Short answers only.
How is this different from grep or ripgrep? ▼
grep finds exact strings. ez-search finds meaning. Search "authentication middleware" and it finds the auth code even when those exact words aren't in the file.
Does it send my code to the cloud? ▼
No. Everything runs locally. Zero network calls. Your code never leaves your device.
What file types does it support? ▼
Code (.ts, .js, .py, .go, .rs, .java, more), docs (.md, .txt, .pdf), and images (.jpg, .png, .gif, .webp, .svg).
How much disk space does it need? ▼
Models are ~500MB total, cached globally across projects. Per-project index is usually a few MB.
Is it free? ▼
Yes. ISC license, 100% open source. No paid tier, no API costs.
grep finds strings. ez-search finds meaning.
Free. Local. Open source.