Six resources for vibe coding with AI

Vibe coding moves fast. These six resources are the ones to check regularly.

MCP directories

MCP (Model Context Protocol) directories list services that give AI agents context and control over outside systems. Check them weekly, and you will see quickly whether a new service fits your workflow. Two directories to check are Glama and Smithery.

Trending GitHub repositories

GitHub's trending page shows projects that are gaining traction, filtered by day or week. Browse it weekly for projects that match your interests or current work.

Clone a project and run it. Tinkering with the code is one of the most effective ways to learn a new technique. A project might also suggest an addition to your AI Power Stack. All you need is a GitHub account.

Rule directories

Rule directories collect rules that steer AI code generation toward a chosen behavior. Good rules encode what experienced engineers do, from breaking a problem apart to following coding standards. Reusing them means you do not start each project from scratch.

To judge a rule, check how many people use it. Adoption is a useful signal of practical value. Not So Bright Ideas shows user adoption counts for its rules, so it is a useful proxy for community validation. Cursor Directory and Playbooks are useful collections too.

Level up your own knowledge

Good tools do not replace understanding. Learn from engineers and researchers who know model performance, benchmarks and the principles behind AI. I follow channels such as @engineerprompt, @ycombinator, @fireship, @samwitteveenai and @IBMTechnology on YouTube.

Applying that knowledge to a real project is harder. workflows.diy offers 1:1 coaching calls for people and teams adopting an AI-first development methodology. I run this service as an AI automation consultant and AI-first software developer. My own workflow is AI-exclusive, and Google Gemini does most of the coding.

Free resources cover the basics of AI. workflows.diy focuses on applying AI to your context and business goals. It helps teams find AI opportunities, build custom AI solutions and set an AI strategy and roadmap.

Boilerplates

Boilerplates are pre-built configurations and project structures that experienced developers write for common project types. Starting from one skips the setup work for authentication, payment integration, database connections and basic UI layouts.

Reinventing these parts for each project wastes time. A boilerplate gives you a tested base that someone else already built, and the options range from free open-source kits to paid, feature-rich ones.

To use one, list the parts that do not set your project apart, such as user authentication, subscription billing or a UI framework setup. Search for boilerplates that cover those parts and match your stack. Build your features and business logic on top of the boilerplate.

Boilerplate List is a good place to find starter kits. It sorts them into web apps, SaaS products, mobile projects and more, and you can filter by stack or cost.

n8n and workflows.diy templates

n8n.io is a low-code automation platform for complex, multi-step workflows. It integrates with many third-party services and places few limits on what you can build. You can add custom code nodes in Python or JavaScript when the built-in nodes do not cover your logic.

n8n suits AI-first development. It supports AI agent chat and lets you scrape and summarize webpages with AI, as in my popular Jina.ai scraper workflow. Its community publishes a large library of n8n community templates, and the platform matches the codability and integrability AI-first toolchains need.

At workflows.diy, we implement AI solutions for organizations, with a focus on n8n. I have shared 33 distinct workflow templates on n8n.io. The most popular are covered in the top n8n AI automation workflows by popularity.

They include AI-powered reporting, automated workflow backups, comparing vision models, monitoring YouTube channels, analyzing YouTube video comments and creating blog content from PDFs. n8n.io itself has over 2000 publicly available templates from the community and the n8n team, plus guides.