Search projects, hackathons, sponsors and tech

    ← All projects

    Creble

    CRE Workflow AI Builder is an AI-powered development platform

    Demo

    About

    CRE Workflow AI Builder is an AI-powered development platform that converts natural language descriptions into production-ready **Chainlink CRE (Chainlink Runtime Environment) workflows**. It's designed to dramatically reduce the complexity of building automation workflows for Chainlink's decentralized oracle networks

    How it's made

    **How It Works** The platform follows a streamlined 4-step workflow: 1. **Natural Language Input** → User describes their workflow need in plain English (e.g., "Fetch ETH price every 5 minutes and store it on-chain") 2. **AI Code Generation** → The backend uses the Groq API (llama-3.3-70b model) powered by specialized system prompts that enforce strict CRE architecture rules. The AI generates valid, production-ready code in TypeScript or Go. 3. **Multi-View Output** → Generated workflows are displayed across three integrated views: - **Code View** – Syntax-highlighted code with copy/download options - **Diagram View** – Visual representation of the workflow's trigger→callback→capabilities→handler flow using React Flow - **Simulation Console** – Local validation and mock execution to test the workflow before deployment 4. **Conversation Persistence** → All generation history is saved to the database (using World ID authentication), allowing users to revisit past workflows and iterate. --- **What Problem It Solves** | Problem | Solution | |---------|----------| | **High barrier to entry** for CRE development | Natural language interface eliminates need to memorize complex SDK APIs | | **Code generation errors** from hallucinated APIs | System prompt enforces strict architecture rules; AI never invents invalid patterns | | **Unclear workflow structure** | Visual diagram auto-generated from code shows the exact execution flow | | **Risk of untested code** | Built-in simulation validates syntax and simulates trigger execution before deployment | | **Lost work history** | Automatic conversation storage preserves all generated workflows | --- **Key Features** - ✅ **Multi-language support** – Generate workflows in TypeScript or Go - ✅ **Strict CRE compliance** – System enforces valid Trigger → Callback → Capabilities → Handler patterns - ✅ **Authentication** – World ID integration for secure user identification - ✅ **Visual debugging** – Workflow diagrams help developers understand execution flow - ✅ **Simulation testing** – Validate workflows locally before deployment - ✅ **Conversation history** – All prompts and generated code saved per user --- **Architecture** ``` Frontend (React + Vite) ├─ HeroPrompt (input interface) ├─ GenerationView (split panel: chat + code output) │ ├─ ChatPanel (AI explanation) │ ├─ CodeOutput (syntax-highlighted code) │ ├─ WorkflowDiagram (visual flow using React Flow) │ └─ SimulationConsole (mock execution) └─ World ID Login (authentication) Backend (Supabase Edge Functions) ├─ generate-workflow (streams AI responses from Groq API) ├─ world-id-verify (validates World ID proofs) └─ Database (stores conversations & user profiles) ```