Skip to main content

​
VibeLearn

VibeLearn is a plugin for Claude Code and Cursor IDE that captures what you build during coding sessions, extracts learning concepts, and generates quiz questions. Run vl quiz after a session to retain what you learned.

​
How It Works

Every time a session ends, VibeLearn automatically runs a 5-step analysis pipeline:
1

Stack Detection

Reads package.json, pyproject.toml, go.mod, etc. to identify your frameworks, ORMs, and tools.
2

Static Analysis

AST-parses files you edited to identify code patterns — custom hooks, API routes, DB queries, auth patterns.
3

Concept Extraction

A single LLM call produces a session narrative and a list of concepts you encountered (e.g., “React Server Actions”, “SQLite WAL mode”, “HMAC signing”).
4

Quiz Generation

A second LLM call generates targeted quiz questions per concept across 7 formats: multiple_choice, code_reading, spot_the_bug, fill_in_blank, open_ended, true_false, and ordering. Question type is selected based on concept difficulty (junior / mid / senior). Open-ended senior answers are evaluated by a 5-dimension rubric.
5

Sync

HMAC-signed sync to vibelearn.dev. Queued offline if unavailable — retried next session.

​
Quick Start

# Install the plugin (Claude Code)
/plugin install vibelearn

# Configure an AI provider (Gemini free tier recommended)
# Add VIBELEARN_GEMINI_API_KEY to ~/.vibelearn/settings.json

# Code normally — after your session ends, quiz yourself:
vl quiz

​
The vl CLI

vl quiz              # Interactive quiz — all pending questions
vl quiz --session    # Quiz from the last session only
vl status            # Sessions analyzed, top concept categories, mastery overview
vl gaps              # Concepts not yet mastered (score < 50%)
vl login <api-key>   # Connect to vibelearn.dev
vl login --status    # Check current auth status

​
Adaptive Difficulty

VibeLearn tracks mastery per concept and adapts quiz difficulty automatically:
  • 3 correct in a row → promoted to the next level (junior → mid → senior)
  • 1 wrong answer → dropped back a level
  • Follow-up questions are inserted mid-quiz when a junior question is answered correctly, to probe deeper understanding
  • Senior open-ended answers are evaluated by a 5-dimension rubric: accuracy, depth, trade-offs, practical reasoning, and completeness

​
Privacy

Wrap any content in <private> tags to prevent storage or sync: 
Please review <private>my-api-key: sk-...</private> this config
Content inside <private> tags is stripped at the hook layer before reaching the worker or database.

​
Architecture

VibeLearn runs as a plugin with 5 lifecycle hooks: 
SessionStart → UserPromptSubmit → PostToolUse → Stop → SessionEnd
The worker service runs locally on port 37778, handles the analysis pipeline asynchronously, and stores all data in SQLite at ~/.vibelearn/vibelearn.db. VibeLearn does not inject context into your prompts. It only observes sessions from the outside and analyzes them after they end.

Installation

Get up and running in 5 minutes

Architecture

How the pipeline works end-to-end

Configuration

AI providers and settings

Cursor Integration

Use VibeLearn with Cursor IDE