Interactive guide · 2026 edition · 80+ tools

AI Coding & Vibe Coding Tools, Visualised

The complete landscape of AI-powered development tools in 2026 — code editors, coding agents, vibe coding app builders, code review, testing, design-to-code, MCP ecosystem, guardrails, and LLM ops. Pick your stage, filter by what matters, and find the tools that fit.

/ search 1-4 stage Esc reset

The landscape at a glance

— · click a node to jump to its card

Which stage are you at?

Pick one — the tools below adapt, and cards fade to only what fits.

All tools

Decision cheat-sheet

Pick by scenario

  • Solo dev, quick prototype: Cursor + v0 + Bolt. Fastest idea-to-app path.
  • CLI power user: Claude Code or Gemini CLI. Terminal-native, agentic, deep codebase understanding.
  • Non-technical founder: Lovable or Replit Agent. Describe what you want, get a working app.
  • Enterprise team: GitHub Copilot Enterprise + CodeRabbit + Augment Code. Compliance, SSO, audit trails.
  • Open-source purist: Zed + Aider + Kilo Code + Qodo. All open source, bring your own LLM.
  • Full autonomy: Devin or Claude Code with Agent Teams. Plan, write, test, deploy end-to-end.

Pick by constraint

  • Zero budget: Gemini CLI (1,000 req/day free) + Zed + Cline. Completely free.
  • Privacy-first: Tabnine (on-prem) + Dyad (local) + Ollama. Nothing leaves your machine.
  • Best code quality: Claude Code (80.8% SWE-bench) + CodeRabbit. Highest accuracy.
  • Fastest output: v0 for UI + Bolt for full-stack. Minutes to working app.
  • Existing IDE: GitHub Copilot (any IDE) or JetBrains AI (JetBrains). Stay in your editor.
  • Team workflow: Copilot Business + CodeRabbit + Mintlify. Integrated into PR flow.

Anti-patterns to avoid

  • Using AI app builders for complex, custom business logic — they're great for MVPs, not enterprise backends.
  • Blindly accepting AI-generated code without review — AI makes confident mistakes.
  • Paying for 5 overlapping AI coding tools — pick one editor + one agent + one review tool.
  • Skipping MCP security — 1,000+ unprotected servers are exposed publicly.
  • Not setting up guardrails for AI-generated code in production — use Snyk or SonarQube.
  • Ignoring AI-generated test quality — AI writes tests that pass but don't catch real bugs.
Tweet Reddit LinkedIn Star on GitHub