auyelbekov

auyelbekov / rawq

Public

Context retrieval engine for AI agents — semantic + lexical search over codebases

ai-agents cli code-search developer-tools embeddings
10
0
100% credibility
Found Mar 19, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

rawq is a standalone offline tool that performs hybrid semantic and keyword search on codebases to retrieve precise snippets for AI agents.

How It Works

1
🔍 Discover rawq

You hear about a handy tool that helps AI assistants quickly find the exact code snippets they need in your projects, without wasting time on irrelevant files.

2
📥 Get it running

With one simple command, you download and install the tool on your computer—it adds itself to your tools and is ready to use right away.

3
📁 Point to your project

You tell the tool about your codebase folder, and it smartly prepares an index of your code so searches are lightning-fast next time.

4
💭 Ask about your code

You type a natural question like 'how does database retry work?' and the tool instantly understands and searches your entire project.

5
✨ See perfect matches

It shows you only the most relevant code chunks with file paths, line numbers, confidence scores, and even syntax highlighting.

🚀 Supercharge your AI

Now your AI agents get precise context without token waste, making them smarter and faster on your codebases—everything works beautifully!

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 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 rawq?

rawq is a Rust-based context retrieval engine for AI agents, delivering hybrid semantic and lexical search across codebases. It indexes your repo offline as a single binary, returning precise code chunks with file paths, line ranges, scopes, and confidence scores—slashing token waste from dumping entire files into LLMs. Run it via CLI like `rawq "database retry" ./src` for instant results in terminal or JSON for agents.

Why is it gaining traction?

Unlike basic grep or full-file stuffers, rawq fuses ONNX embeddings with BM25 for smarter context retrieval, plus git-diff scoping and auto-reindexing on changes. Agent devs love the streaming NDJSON, token budgets, and daemon mode keeping models hot; GPU accel via CUDA/CoreML/DirectML makes it fly on laptops. Fully offline RAG beats cloud-dependent tools, with tree-sitter chunking for 16+ languages.

Who should use this?

AI agent builders integrating context retrieval RAG with Anthropic Claude or GitHub Copilot workflows—think backend devs querying retry logic in microservices, or frontend teams pulling React hooks. Perfect for GitHub Actions automating context github copilot prompts, or LLM chains needing precise codebase cues without token bloat. Skip if you're not doing agentic flows.

Verdict

Grab it if you're prototyping context retrieval agents—installs in one curl, works out-of-box with snowflake-arctic-embed. At 10 stars and 1.0% credibility, it's early (light tests, beta vibes), but solid docs and MIT license make it worth a spin for offline RAG benchmarks. Rebuild indexes post-changes until watch mode matures.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.