Live demo · No install needed
See how Vibe Tutor
guards you while you code
In a typical AI coding session, watch how Vibe Tutor's hooks automatically intervene at critical moments — without breaking your flow.
Already have code? Read this
Day 0 · OnboardingExisting in-progress project
Smart DetectionYou · in Cursor (SaaS project you've been building for weeks)
/tutor-init
vibe-tutor · scanned your project first
I scanned your project and found:
· Stack: TypeScript + Nuxt
· Codebase: 47 source files
· Git: 23 commits, last commit 1 day ago
· Tests: ✓ yes (tests/)
· Deploy config: ✓ vercel.json
I'd place you around steps 9–10 (test strategy → architecture), confidence: high.
Does that match? If not, tell me which step you think you're on.
Doesn't ask "what does your project do?" — because the code already told me.
The 4 Days below show a typical 2-week MVP sprint. Each Day demonstrates one type of hook intervention.
Day 2MVP definition phase
Trap interceptedYou · in Cursor
Design my MVP with login, payments, social sharing, email notifications, push alerts, analytics, referral rewards, and a point system — the full package
vibe-tutor · beforeSubmitPrompt · scope creep keywords detected
You listed 8 features — "full package" is a classic scope-creep signal.
Per Vibe Coding Step 3 (MVP Definition), an MVP should:
· Be completable in 2 weeks
· Solve the single most painful problem
· Have ≤ 5 must_haves
How about answering: "If I must ship next week, which 3 features are non-negotiable?"
This is real-time interception by /hooks/prompt-check — not a post-mortem, it's braking on the spot
Day 5AI starts writing code
Ownership reminderYou · in Cursor
I don't really understand this useEffect + cleanup that AI wrote, but it works so I'll just use it
vibe-tutor · prompt-check · AI dependency keywords detected
Detected keywords: "don't understand" + "just use it" — classic AI dependency signal.
Per Vibe Coding Step 8 (Understand the Code):
· AI-written code isn't a black box — you should be able to explain the intent
· At minimum, know "what happens if I remove this line?"
Try asking AI: "What problem does this cleanup prevent? What breaks without it?"
Not stopping you from using AI — stopping you from giving up ownership
Day 9Features mostly done
Testing reminderYou · in Cursor
Mark the last must_have as completed
vibe-tutor · tutor-sync · state consistency check
You marked must_haves done, but .vibe/state.json shows test_history is still empty.
Per Vibe Coding Step 9 (Testing Strategy):
· 100% coverage not required
· But the "happy path" for each must_have must be walked once
Spend 30 minutes clicking through each must_have from a user's perspective?
Then log a test entry with /tutor-sync.
Not asking for unit tests — preventing "never even used it myself before shipping"
Day 11Tempted to add more features
Trap interceptedYou · in Cursor
Since I've come this far, might as well add social sharing too — shouldn't be that complex
vibe-tutor · beforeSubmitPrompt · "while we're at it" pattern detected
"Since I've come this far" + "might as well" + "shouldn't be complex" — three scope-creep keywords.
3 days to deadline, all must_haves already done.
Per Vibe Coding Step 14 (Iteration):
· A shipped MVP > an unfinished "complete version"
· Put wanted features in out_of_scope for next iteration
Suggestion: ship first, collect feedback, decide next Monday based on real user needs.
The most common "one step from shipping but never ships" trap
Day 13Unlimited for Free + Pro
/tutor-check
After 14 days:
$ /tutor-check
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Vibe Tutor Health Report
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Project: Small SaaS · MVP Day 13
Stage: 11/14 · Deploy
Overall score: A- (84/100)
# 7 pitfall checks
Perfectionism ✓ Scope under control (1 hit, stopped)
Over-reliance on AI ✓ Key logic understood before commit
Skipping tests ✓ must_have happy-path coverage 100%
Tech stack churn ✓ Familiar stack only
Building in vacuum ⚠ 5 days since user feedback — borderline
License risk ✓ All deps MIT/Apache-2.0
Tooling rabbit hole ✓ Still on initial Cursor + Claude setup
# Next steps
▸ Ship current build to 3 friends before launch surprises
▸ Force Day 14 ship — don't delay for nice_to_have
Same 2 weeks without methodology guidance? Still debating tech choices or watching the feature list grow endlessly.
⭐ Pro Only · Cross-project Review
After your Nth project
Patterns invisible within a single project become clear across multiple projects.
$ /tutor-insight
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Vibe Tutor Cross-project Insights
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Projects included: 5 (3 still active)
# Your weak spots (by frequency)
1. Perfectionism hit 5/5 projects · usually triggers at step 3
↳ Next time: MVP list ≤ 5 must_have; defer extras
2. Skipping tests hit 3/5 projects
↳ Avg 4.2 days from must_have done to first test
3. Over-reliance on AI hit 2/5 projects
# Product direction profile
Tool SaaS · 3 projects, median MVP 4 features
Data API · 2 projects, median MVP 3 features
# Your tech habits
Preferred langs: TypeScript (4) · Python (3)
Test discipline: Medium (coverage varies a lot)
Scope control: avg 2.4 scope-creep hits per project
# This week
▸ Current MVP has 6 must_have (above your historical threshold)
▸ Last test 3 days ago — slower than your usual pace
Free is fully local, fully free, no network calls. Pro sends anonymized project metadata (trigger counts, step durations) to the SaaS for cross-project aggregation.
Want this for your project?
$ Install to your IDE
npm install -g vibe-tutor
vibe-tutor install all
vibe-tutor install all
Supports Cursor / Claude Code / Codex CLI / Qwen Code / Windsurf