Interactive sandbox for exploring AI agent workflows
Playground · real capture
02 · Project brief
About this project
Role
Tool · self-directed
Type
Web App
Status
Building
Category
tools
A personal learning project for experimenting with multi-agent AI interactions. Features 12 specialized agents powered by Claude (Opus 4.5, Sonnet 4.5, Haiku 4.5), with personas ranging from UX researchers to backend developers. Built to explore roleplay-based system prompts, agent hand-offs, and conversation context preservation.
03 · Capabilities
06 capabilities
What I built
01
12 specialized AI agents with distinct personas
02
Real Claude API integration (Opus 4.5, Sonnet 4.5, Haiku 4.5)
03
Multi-turn conversations with context preservation
04
Agent hand-off capabilities
05
XP and mastery tracking gamification
06
Roleplay-based personality system
04 · Product screens
03 views
Inside the build
05 · Build details
Technology & access
Technology
Next.js 15
React 19
TypeScript
Claude API
Zustand
Tailwind CSS
06 · Contact
Got a product that deserves to be ready?
One conversation is enough to know whether this is a fit. Tell me what needs to work better.
