How to vibe codeMongoDB

Prerequisites

• Node.jsfree

  Required to run the lightweight TypeScript subset mock server.

• GitHubfree

  Version control for source code.

AI coding tools

• Claude Code

  Best-in-class multi-file agentic coding tool for bootstrapping a toy document store subset.

• Cursor

  AI-native editor for inspecting and debugging storage engine code.

Build guide

1. 01Project Scaffolding and Document Data Model

   Initialize a TypeScript project with Next.js and define the schema structures for representing JSON-like documents and collections.

   Scaffold a new Next.js project with TypeScript and Tailwind CSS. Create a database abstraction layer using Turso/SQLite that accepts arbitrary JSON objects and stores them inside a serialized text column representing a MongoDB collection. Implement basic collection creation and document insertion endpoints adhering to a simple CRUD interface.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

2. 02Query Filter Parser and Projection Engine

   Build a mini query parser to handle basic MongoDB-style query operators like $eq, $gt, $in, and $regex against stored JSON documents.

   Write a query parser module in TypeScript that interprets subset MongoDB query operators ($eq, $gt, $lt, $in, $ne). Given an array of JSON documents stored in memory or fetched from SQLite, evaluate the filter predicates recursively against nested object properties using dot notation paths (e.g., 'address.city'). Return matching documents.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

3. 03Basic Aggregation Pipeline Stages

   Implement a pipeline executor supporting $match, $group, $sort, and $project stages to mimic MongoDB's aggregation framework.

   Implement an aggregation pipeline runner in TypeScript that processes an array of documents sequentially through a list of stages. Support $match (filtering using the query parser from step 2), $group (aggregating by an id expression with $sum or $avg accumulators), $sort, and $project (reshaping output fields). Write unit tests verifying pipeline execution correctness.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

4. 04Vector Embedding Storage and Similarity Search

   Add support for storing vector embeddings alongside documents and perform cosine similarity calculations for semantic search.

   Add a vector search capability to the toy document store. Allow documents to hold an embedding array field. Implement a cosine similarity function in TypeScript that compares a query vector against all document vectors in a collection, sorts by similarity score, and returns the top matching documents with their scores.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

5. 05Graphical Management Dashboard UI

   Build a web-based UI mimicking MongoDB Compass to inspect collections, view documents, run query filters, and visualize aggregation pipelines.

   Build a management dashboard UI in Next.js resembling MongoDB Compass. Include a sidebar listing collections, a central document browser table with JSON tree view inspection, an editor tab for running filter queries and aggregation pipelines, and a form to insert new JSON documents. Connect the UI components to the API routes built in previous steps.

   ⧉copy promptopen in Claude CodeCopen in Codexopen in Cursor

Cost vs paying for MongoDB