datadictionary
A comprehensive, interactive data dictionary and MCP server for the Kwikly dental staffing platform, documenting 154 Rails models, business metrics, API endpoints, and domain knowledge.
4 commits over 2 days
What it is
Built to provide developers and AI assistants with instant access to the Kwikly platform's data structures, business logic, and domain knowledge. The tool includes a Next.js web app that visualizes 154 Rails models with associations, validations, and methods, plus a Model Context Protocol (MCP) server for AI integration. It enables natural-language SQL generation, model exploration, and business context retrieval. The project also uncovered critical inconsistencies in cancellation tracking across the codebase.
Why I built it
Likely created to solve onboarding and knowledge-sharing challenges in a complex multi-tenant SaaS codebase. The comprehensive static analysis of Rails models, API endpoints, and business metrics suggests a need to document tribal knowledge and enable faster developer ramp-up. The MCP server integration indicates an interest in leveraging AI assistants for development tasks. The discovery of cancellation tracking issues points to a hidden audit/cleanup motivation.
Tech stack
Next.js 15 with TypeScript and App Router, Tailwind CSS, Heroicons, Headless UI. Static JSON data files (no runtime database). Ruby and PHP extraction tools for analyzing the Rails codebase. Custom MCP (Model Context Protocol) server with JSON-RPC 2.0 for AI integration. Deployed to Vercel.
Timeline & effort
Last active Oct 14, 2025 — 4 commits across 2 active days (spanning 10 days).
Built over 11 calendar days with 2 distinct active days and 4 commits (Oct 3–14, 2025). The commit pattern suggests intensive bursts of work, likely driven by exploratory static analysis followed by web app implementation. The deliverable is substantial—154 models, 1,000+ methods, 462 user perspectives, and a full MCP server—indicating focused, possibly AI-assisted development sessions.
Did it turn out useful?
The project shipped to production on Vercel and delivers concrete value: it documents a complex Rails codebase, provides searchable interfaces for models/APIs/metrics, and integrates with AI assistants via MCP. The discovery of critical cancellation tracking issues demonstrates real analytical utility. The comprehensive documentation and working MCP server suggest ongoing use by the team, positioning this as an internal developer tool rather than a one-off experiment.