Abhay Singh Rathore Logo
Let's talkBook Intro Call →
← Back to Work
Snippet Hive Hero

Snippet Hive.

A public library for AI-generated code snippets, built in 6 weeks during Buildspace.

RoleFull-stack / Co-founder
Timeline6 weeks
StackReact · Node · MongoDB · Chrome & VS Code APIs
StatusSunset

The Problem.

Developers using AI often spend 10–20 minutes regenerating the same complex blocks (authentication flows, hero sections, boilerplate) because there is no shared public library to instantly grab perfected AI outputs.

My Role.

  • → Built the core web platform and snippet repository
  • → Developed the VS Code extension for direct IDE integration
  • → Created the Chrome extension for AI chat integration
  • → Implemented a community rating and upvote system

The Solution.

Snippet Hive Web App

A community-driven repository where developers can share, rate, and instantly reuse code. Instead of just a website, it integrates directly into the editor with a VS Code extension and into AI chats with a Chrome extension.

VS Code Extension
Chrome Extension

Technical Decisions.

Why Next.js?

→ Server-side rendering for the marketing site, static export for the dashboard. Fastest path to production without splitting repos.

Why MongoDB over Postgres?

→ Snippet data is document-like: tags, metadata, nested fields. Mongo gave us flexibility early. We'd reconsider at scale.

Auth strategy?

→ JWT with refresh tokens. Considered NextAuth but wanted more control over session behavior for the VS Code extension context.

State management?

→ Zustand over Redux. Simpler API, less boilerplate, good enough for this scope.

What Broke.

The sync engine was the hardest part. We underestimated how many edge cases come from offline edits: conflicts, ordering, partial syncs. We had to rebuild it twice. The second time, we started with the conflict resolution logic first instead of the happy path. That mindset shift changed how I approach any stateful system now.

Outcome.

  • → Still early: here's what the first 10 users said.
  • → 100% faster snippet retrieval vs previous workflow (self-reported)

What I'd Do Differently.

"If I rebuilt this today, I'd focus entirely on semantic search using AI embeddings rather than manual tagging. The real friction wasn't storage; it was retrieval. Having an AI perfectly understand the context to surface the exact snippet instantly would have matched the seamless experience of just regenerating it."

Next up.

Enjoyed this project? Check out my other work or let's build something great together.