rjyo

rjyo / memory-search

Public

Standalone memory search with hybrid search (vector + BM25) for Claude Code

22
0
100% credibility
Found Feb 09, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

memory-search is a hybrid search system that indexes and retrieves content from project memory files (MEMORY.md and memory/*.md) for AI coding agents using vector similarity and keyword matching.

How It Works

1
📰 Discover memory helper

You hear about a handy tool that lets your AI coding assistant remember your project notes, decisions, and preferences so it stays consistent.

2
🔧 Add to your workspace

You easily add this memory tool to your coding agent's toolkit, making it ready to use right away.

3
📝 Create your memory notes

In your project folder, make a simple notes file called MEMORY.md and start writing down key ideas, choices, and daily thoughts in a memory folder.

4
💾 Save new memories

Chat with your AI assistant using phrases like 'remember this preference' or '/memory remember...', and it stores them safely in your notes.

5
🔍 Recall past decisions

Ask 'what did we decide about login?' or similar, and your assistant instantly pulls up the exact snippets from your memories with confidence scores.

🎉 AI remembers everything

Now your coding sessions feel seamless as the AI always recalls your project's history, preferences, and details without you repeating yourself.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 22 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 memory-search?

Memory-search is a TypeScript tool that builds a hybrid search engine (vector similarity + BM25 keywords) over your project's MEMORY.md and memory/*.md files, letting AI coding agents like Claude Code or Cursor recall past decisions, preferences, and notes. Install it standalone via `bunx skills add rjyo/memory-search`, sync with `bunx memory-search --sync`, and query via `/memory "auth decisions"` or programmatic API. It runs embedded or standalone with local embeddings (no API key) or OpenAI fallback, using Bun and SQLite for lightweight storage—ideal for searching memory without gdb-like hacks or external services.

Why is it gaining traction?

Unlike basic keyword tools or full RAG stacks, it blends semantic and exact-match search out-of-the-box, with auto-context injection for Claude hooks and CLI scripts for quick sync/search. Developers dig the zero-setup local model (300MB download) versus cloud-dependent alternatives, plus seamless integration with 30+ agents—no more repeating "what did we decide?" in every session. It's a practical memory searcher for GitHub workflows, standing out in standalone GitHub repos like brave or selenium standalone setups.

Who should use this?

AI-assisted coders on Claude Code, Cursor, or Codex who juggle long projects and hate context loss between sessions. Perfect for solo devs tracking architecture choices in MEMORY.md, or teams dumping daily notes into memory/ folders for hybrid recall. Skip if you're not in TypeScript/Bun ecosystems or prefer embedded vs standalone memory in tools like Citra/Cemu modders.

Verdict

Early-stage with 19 stars and 1.0% credibility—docs are solid, tests exist, but low adoption means watch for bugs. Try it if you're building AI coding flows; it's a clever standalone memory search GitHub gem worth forking for custom tasks.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.