Dicklesworthstone

Dicklesworthstone / frankensearch

Public

Two-tier hybrid search for Rust: sub-millisecond initial results via potion-128M, quality-refined rankings in 150ms via MiniLM-L6-v2. Combines lexical (Tantivy BM25) and semantic (vector cosine) search with Reciprocal Rank Fusion. Progressive iterator API, f16 SIMD vector index, feature-gated compilation.

embeddings full-text-search hybrid-search information-retrieval nlp
37
8
100% credibility
Found Feb 23, 2026 at 31 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

frankensearch is a standalone command-line tool and Rust library for fast hybrid local search over codebases and documents, delivering quick initial keyword and semantic matches followed by refined high-quality rankings.

How It Works

1
🔍 Discover fast local search

You hear about a tool that searches your project files super quickly, finding answers to questions like 'how does retry backoff work?' without waiting.

2
📦 Install with one command

Run a simple copy-paste command to get the fsfs search tool ready on your computer in seconds.

3
📁 Point it at your project

Tell it to scan your folder, and it builds a smart index of your files automatically.

4
❓ Ask natural questions

Type a question about your code, like 'structured concurrency', and hit enter.

5
⚡ Get instant smart results

See top matches right away from keywords and meaning, then watch them get even better as it refines.

✅ Find exactly what you need

Your searches now reveal relevant files fast and accurately, making you super productive in your projects.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 31 to 37 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is frankensearch?

Frankensearch is a Rust library and CLI (fsfs) for local hybrid search on codebases or docs. It delivers sub-millisecond initial results blending lexical BM25 and fast semantic cosine similarity, then refines rankings in 150ms using reciprocal rank fusion for better quality. Developers get a progressive iterator API for streaming hits, plus an index with f16 quantization and feature-gated compilation for custom builds.

Why is it gaining traction?

It combines lexical precision with semantic recall in a fully local setup—no cloud dependency—while the two-tier design ensures responsive UIs with fast initial hits. The CLI supports watch-mode indexing, JSONL streaming for APIs and agents, and explainable results, outpacing grep for intent-based queries. Feature-gated deps keep binaries lean, appealing to embedded use cases.

Who should use this?

Rust devs building local search into CLI tools, IDE plugins, or RAG pipelines for code/docs. AI agent builders needing hybrid retrieval on private repos. Local-first app makers frustrated with ripgrep's keyword limits but wary of heavy vector DBs.

Verdict

Try it for Rust-native hybrid search prototypes—docs, benchmarks, and installer shine despite 23 stars and 1.0% credibility signaling early maturity. Skip for production scale until more adoption.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.