cursorrules
01Cursor rule Cursor
You are an expert AI Coding Agent assisting technical workshop facilitators.
Cursor rule Cursor
You are an expert AI Coding Agent assisting technical workshop facilitators.
Instructions file CodexOpenCode
Instructions for JAICHANGPARK/workshop-harness, covering agents.md - agent instructions for workshop harness, 🎯 repository overview & purpose, ⚡ quick agent commands & workflows (under-the-hood), 1. one-click full workshop generation across skills and 2. android hands-on workshop initialization.
Instructions file
Instructions for JAICHANGPARK/workshop-harness, covering 🎯 repository overview & claude code role, 🤖 critical rules for claude code, 📁 20 specialized agent skills index and 🔗 references.
Skill Claude CodeCodex
Builds hands-on workshops on A2UI (Agent-to-User Interface) protocol and Flutter GenUI (Generative UI). Teaches developers how to construct dynamic, ephemeral UI surfaces streamed directly from AI agents using declarative JSON schemas and type-safe WidgetCatalogs.
Skill Claude CodeCodex
Builds multi-language (Python, TypeScript, Go, Kotlin) autonomous agent and multi-agent system workshops using Google Agent Development Kit (ADK). Generates Coordinator-Worker agent architectures, tool bindings, Human-in-the-Loop (HITL) checkpoints, and trajectory evaluation labs.
Skill Claude CodeCodex
Builds modern Android Generative AI and on-device AI hands-on workshops using Jetpack Compose, Kotlin 2.x, Gradle Kotlin DSL, the official Google GenAI Kotlin Android SDK (com.google.genai), Google Android CLI (android), and official Android Agent Skills. Enforces reactive ViewModel architecture, streaming responses…
Skill Claude CodeCodex
Generates Google Colab ready interactive notebooks (.ipynb) from workshop labs, injects zero-setup runtime cells (pip dependencies, GPU check, Colab Secrets / API keys), creates 'Open in Colab' badges, and automates headless execution & smoke testing via the Google Colab CLI (colab-cli).
Skill Claude CodeCodex
Audits cross-architecture compatibility across Apple Silicon (arm64), Intel Mac (x8664), Windows (x86/ARM64), Linux, and ChromeOS. Identifies known risks for tools like LM Studio, Docker, and MLX, and generates a fallback matrix document (docs/00-architecture-compatibility-matrix.md).
Skill Claude CodeCodex
Translates complex technical concepts, AI model mechanics, and architecture errors into intuitive ELI5 (Explain Like I'm 5) real-world analogies, visual mental maps, and beginner-friendly cheat sheets for workshop attendees and facilitators.
Skill Claude CodeCodex
Explain any topic, code logic, system architecture, or technical error like I'm 5 years old (ELI5) using simple physical analogies, minimal text, and visual diagrams/artifacts.
Skill Claude CodeCodex
Builds cross-platform Flutter Generative AI hands-on workshops using Flutter 3.x, Dart 3.x, the official googlegenerativeai package, official Flutter Agent Plugins (flutter/agent-plugins), and Dart Package Skills. Enforces Material 3 design, reactive state management, streaming chat responses, Dart/Flutter MCP server…
Skill Claude CodeCodex
Builds step-by-step hands-on lab curriculum (Lab 1, Lab 2, Lab 3), starter and final code templates, prompt packs with Output Schema specifications, and enforces a mandatory references section in all generated documents.
Skill Claude CodeCodex
Analyzes terminal error logs in under 10 seconds and outputs actionable fix commands. Includes API Key security scanning protocol to prevent accidental credential exposure.
Skill Claude CodeCodex
Integrates workshop content created by workshop-harness with Open Codelabs platform. Automates converting workshop labs into Open Codelabs manifest format (codelab.yaml), generating step-by-step markdown, mapping starter/final code workspaces, attaching prerequisites & materials, and publishing via oc CLI, REST API…
Skill Claude CodeCodex
Converts markdown documentation files into publication-ready PDF handouts and preview contact sheets using ReportLab, PyMuPDF (fitz), and Pillow.
Skill Claude CodeCodex
Generates OS-specific (macOS Apple Silicon/Intel, Windows PowerShell/WSL2, Linux, ChromeOS) prerequisite setup guides covering language/SDK installation (Python, Node.js, Go, Flutter, Rust, Docker), local LLM server API port configuration (Ollama 11434 vs LM Studio 1234), latest AI model downloads, Google AI Studio /…
Skill Claude CodeCodex
Auto-generates attendee FAQ documents (FAQ.md / docs/20-faq.md) tailored to the workshop topic, tech stack, and target audience level (beginner/intermediate) covering hardware, network, and code-related questions.
Skill Claude CodeCodex
Evaluates and audits workshop curriculum, lab guides, starter/final code, and setup documents through 4 distinct attendee reviewer personas (Non-Coder, Novice, Intermediate, Senior Engineer). Generates a multi-persona evaluation report at docs/00-persona-loop-review-report.md using Loop Engineering.
Skill Claude CodeCodex
Generates a minute-by-minute facilitator and TA timeline runbook (RUNBOOK.md), including opening scripts, slide sync markers, TA response time rules, and progress checkpoint curl commands.
Skill Claude CodeCodex
Scaffolds a standard workshop repository structure including docs/, workshop/, prompt-pack/, scripts/, and output/ directories with pre-configured templates and automation scripts.
Skill Claude CodeCodex
Generates high-impact, professional workshop presentation slide decks in Marp Markdown (slides.md) and standalone interactive HTML format (index.html). Syncs 1:1 with facilitator RUNBOOK.md timeline markers, hands-on lab steps, and 'Open in Colab' links.
Skill Claude CodeCodex
Runs automated smoke tests on starter and final code, validates markdown document integrity (broken relative paths, missing image links), and generates a test report.
Skill Claude CodeCodex
Generates OS-specific and RAM-tier-specific (8GB, 16GB, 32GB+) troubleshooting guides and FAQ documents covering offline/network failure scenarios, model download failures, API port mismatches (Ollama 11434 vs LM Studio 1234), and GPU/CPU fallback strategies.
Skill Claude CodeCodex
Performs live web searches to verify the latest release versions and tags for tools (Ollama, LM Studio, uv, Docker) and AI models (Gemini, Gemma 4, Claude, Llama, etc.), ensuring data currency and preventing hardcoded legacy versions in generated documentation.