By rizukirr
Enforce hands-on coding by blocking AI writes to project files, turning it into a tutor that guides top-down learning via chat. Generate calibrated challenges with testable criteria, hints, and reviews. Reference projects, toggle modes, and temporarily allow scoped tasks like adding .gitignore.
npx claudepluginhub rizukirr/no-vibe --plugin no-vibeOne-shot escape hatch — let AI write code to project files for a single task, then restore no-vibe mode
Get a coding challenge based on your project or current no-vibe session
Enter no-vibe mode. AI will not write code to your project files; it will guide you through writing the code yourself, top-down from a high-level API to its foundations, optionally grounded in real reference projects.
One-shot escape hatch to temporarily disable no-vibe write guard for a single scoped task, then restore the marker.
Generate a one-shot coding challenge based on current no-vibe session or project context. User writes all code; AI only gives constraints, hints, and review.
Use ONLY when `.no-vibe/active` marker exists at the project root, or the user has just invoked `/no-vibe` / `/no-vibe on`. Do NOT trigger merely because the user wants to learn or type code themselves without those signals — the marker or explicit command is the required gate. Once active, EVERY reply must begin with the header `[no-vibe] Phase: <0|1a|1b|1c|2|3|4|5|6> · Session: <slug-or-none> · Layer: <n/total-or--> · Next: <action-verb-clause>` and follow the per-turn order: read sessions/<slug>.json → run pre-turn audit → emit header → act (chat-only, no project writes) → log per data-logging.md triggers → update session JSON if state changed. Full contract in SKILL.md "Turn Response Contract" section.
Turn your AI assistant into a personal tutor that empowers you to write the code yourself. It shows you what to build and explains the 'why' while reviewing your work. The plugin adapts to your level and improves its teaching based on your learning progress, all without touching your project files. Keep the muscle memory, lose the dependency.
We’ve all felt it: the more we 'vibe-code', the more our skills slip away. I built this to stop the drain. The AI acts as your mentor—planning and reviewing every step—but you’re the one typing every character. It’s the productivity of AI with the growth of manual coding. Break the dependency loop.
Maximize your workflow by pairing no-vibe with vibekit. Use the vibekit pipeline to build fast, and switch to no-vibe when you want the AI to tutor you through the manual coding process. Stay productive, keep learning
--ref <url> and the AI quotes actual source with file:line citations instead of inventing APIs..no-vibe/**. Codex/Gemini enforce the same rule via instruction.Learner data is local: project-level .no-vibe/data/, global ~/.no-vibe/.
/plugin marketplace add rizukirr/no-vibe
/plugin install no-vibe@no-vibe
Restart Claude Code.
Paste into the relevant CLI:
Fetch and follow instructions from https://raw.githubusercontent.com/rizukirr/no-vibe/refs/heads/main/INSTALL.opencode.md
(Swap INSTALL.opencode.md for INSTALL.codex.md, INSTALL.gemini.md, or INSTALL.pi.md as appropriate. Manual install: see each file directly.)
/no-vibe build a linear layer like pytorch's
Codex uses $ instead of /.
| Command | Effect |
|---|---|
/no-vibe on / off | persistent mode toggle |
/no-vibe <topic> | one-shot lesson |
/no-vibe --ref <url> <topic> | attach a reference project |
/no-vibe --mode concept|skill|debug <topic> | set teaching style |
/no-vibe-btw <task> | one-shot escape hatch — AI may write for this task only |
/no-vibe:challenge [<focus>] | get a coding challenge |
Flags combine: /no-vibe --ref pytorch --mode concept how does autograd work.
| Mode | Best for | Style |
|---|---|---|
| concept (default) | "teach me how X works" | more prose, deeper check-ins |
| skill | "I want to practice writing Y" | muscle-memory repetition |
| debug | "why does Z behave like this" | start from symptom, descend |
| Feature | Claude Code | OpenCode | Pi | Codex | Gemini CLI |
|---|---|---|---|---|---|
| File-write guard (hook) | ✓ | ✓ | ✓ | soft | soft |
| Bash-write guard (hook) | ✓ | ✓ | ✓ | soft | soft |
| Status + resume hint | ✓ | ✓ | ✓ | soft | soft |
| Commands | ✓ | ✓ | ✓ | ✓ | ✓ |
| Skill + learner tracking | ✓ | ✓ | ✓ | ✓ | ✓ |
"soft" = instruction-enforced (no hook surface available); the rule still binds.
Issues and PRs welcome at issues.
MIT
Science-based learning exercises for deliberate skill development during AI-assisted coding
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Share bugs, ideas, or general feedback.
Maintain programming skills by ensuring humans write a minimum percentage of code
In-context coding tutor for Claude Code. Learn from your real project with explanations, quizzes, diagnostics, and belt-based progression — locally and privately.
Personalized coding tutorials that use your actual codebase for examples with spaced repetition quizzes
Keep your coding skills sharp! Periodic manual coding challenges with gamification (streaks, achievements, skip system)
AI-powered Socratic learning mode - Transform Claude into a patient coding mentor that guides you through problem-solving without giving direct answers