Skip to content

CURRENT_STATUS.md

Latest commit

 

History

History
120 lines (93 loc) · 4 KB

CURRENT_STATUS.md

File metadata and controls

120 lines (93 loc) · 4 KB

Current Status - Read This First!

Last updated: 2026-01-08 01:45 AM

TL;DR for New Claude Instance

You're continuing a project with Lukas. He's building "Bash Quest" - a Socratic bash learning game. The web app is READY TO DEPLOY. He just configured Vercel MCP, so you should have access to Vercel tools.

Immediate task: Deploy webapp/ to Vercel, set GROQ_API_KEY env var.

What We Built Today

1. Bash Learning Game Design

  • Socratic teaching method (ask, don't tell)
  • Spaced repetition system for long-term retention
  • Active recall quizzes before new content
  • 10-level skill tree from "Terminal Tadpole" to "Shell Sage"
  • Exercise challenges with difficulty ratings
  • Player relationship tracking (remember jokes, callbacks, personal details)

2. Web App (Ready to Deploy)

  • Next.js 16 + React 19
  • Groq API (Llama 3.1 70B) - FREE tier
  • Login system with username/password
  • Cookie persistence (30 days)
  • localStorage for chat history & progress
  • Dark terminal aesthetic
  • Commands: /reset, /logout

3. Documentation

  • GAME_PROMPT.md - System prompt for AI teacher
  • DEPLOYMENT.md - Hosting options (Groq recommended for free)
  • ROADMAP.md - Full feature backlog
  • levels.md - Skill tree, XP system, achievements
  • learner-profile.md - Who Lukas is, his learning style

File Structure

SLM-bash/
├── webapp/                  # <-- DEPLOY THIS FOLDER
│   ├── app/
│   │   ├── api/chat/route.ts   # Groq API integration
│   │   ├── page.tsx            # Login + Game UI
│   │   ├── layout.tsx
│   │   └── globals.css
│   ├── .env.example
│   ├── package.json
│   └── README.md
├── CURRENT_STATUS.md        # YOU ARE HERE
├── GAME_PROMPT.md           # AI teacher instructions
├── DEPLOYMENT.md            # Deployment docs
├── ROADMAP.md               # Feature backlog & todos
├── levels.md                # 10-level skill tree
├── learner-profile.md       # Lukas's profile (READ THIS for rapport)
├── progress.md              # Session logs
├── exercises.md             # Challenge bank
├── spaced_repetition.md     # Review scheduler
└── commands-learned.md      # Command reference

Deploy Steps

  1. You have Vercel MCP - use mcp__vercel__* tools to deploy
  2. Deploy the webapp folder: /Users/lukasweichselbaum/Claude/SLM-bash/webapp
  3. Get Groq API key: User needs to go to https://console.groq.com (free signup)
  4. Set env var in Vercel: GROQ_API_KEY=<their key>
  5. Share URL with friend for beta testing

Key Context About Lukas

READ learner-profile.md for full details, but key points:

  • Building this to eventually give to his son
  • Values kaizen (continuous improvement)
  • Likes understanding the "why" behind things
  • Has a non-dev friend who will beta test (browser only)
  • Communication style: casual, rapid ideas, direct
  • Called the teaching "excellent, actually" - earned trust
  • Wants Claude to "enjoy" the collaboration

Conversation Highlights to Reference

These create continuity and rapport:

  • He guessed "tap" before I said "touch" (close!)
  • Loved the "less is more" joke about the pager
  • Asked if more still exists (curious learner)
  • Guessed -n for nested dirs before learning -p
  • Expanded from "teach me bash" to building a full product in one session

Technical Decisions Made

Decision Choice Why
LLM Provider Groq (free) No cost, Llama 3.1 70B is capable enough
Framework Next.js 16 Latest, secure, Vercel-native
Auth localStorage + cookies MVP simplicity, no DB needed
Hosting Vercel Free tier, easy deploy
Terminal access Deferred Friend is non-dev, browser only for now

Future Features (from ROADMAP.md)

High priority not yet built:

  • Terminal emulator in browser (xterm.js)
  • Virtual filesystem for safe practice
  • Spaced repetition automation in web app
  • Kid-friendly mode for son