JSPM

@opensvm/vibedev

0.4.0
  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 8
  • Score
    100M100P100Q34881F
  • License MIT

Analyze AI coding assistant usage patterns, costs, and productivity

Package Exports

    This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@opensvm/vibedev) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

    Readme

    vibedev

    cargo install vibedev

    Analyze your AI coding assistant usage patterns, costs, and productivity.

    License: MIT Rust

    Features

    • Multi-Tool Support - Claude Code, Cursor, Cline, Kiro, Copilot, Windsurf, and 15+ more
    • Interactive TUI - Real-time visualization of your AI usage (like dust for logs)
    • Embedded LLM - Offline AI chat to analyze your data (no API keys needed)
    • Deep Insights - Usage patterns, cost estimates, productivity metrics
    • Dataset Export - Create sanitized training datasets from your conversations
    • Backup & Restore - Compress and archive your AI learning data

    Quick Start

    # Install
cargo install --path .

# Discover AI tool logs on your system
vibedev discover

# Interactive TUI dashboard
vibedev tui

# Generate analysis report
vibedev analyze --format html --output report.html

    Commands

    Command Description
    discover Scan system for AI tool logs
    analyze Generate analysis reports (text/json/html/markdown)
    tui Interactive terminal dashboard
    chat Chat with your data using embedded LLM
    daemon Keep LLM loaded for fast queries
    models Manage offline LLM models
    claude Manage Claude Code provider configuration
    backup Create compressed backup archive
    prepare Export sanitized training dataset
    insights Comprehensive HTML dashboard
    compare Compare usage across tools

    Claude Provider Management

    Switch between different API providers for Claude Code:

    # List all supported providers
vibedev claude list

# Show current configuration
vibedev claude show

# Set a provider (saves configuration)
vibedev claude set z.ai --api-key sk-xxx
vibedev claude set openrouter --api-key sk-or-xxx
vibedev claude set chatgpt --api-key sk-xxx
vibedev claude set litellm --api-key sk-xxx

# Apply configuration to Claude Code (updates config files)
vibedev claude set openrouter --api-key sk-or-xxx --apply

# Custom provider with custom endpoint
vibedev claude set custom --endpoint https://api.example.com/v1 --model gpt-4 --api-key xxx --apply

    Supported Providers:

    • z.ai - High-performance Claude API proxy
    • openrouter - Unified LLM API with multiple models
    • chatgpt - OpenAI GPT-4 and other OpenAI models
    • litellm - Local proxy for multiple providers
    • custom - Your own custom endpoint

    Embedded LLM (Offline)

    Analyze your data with a fully offline LLM - no API keys or internet required:

    # Download a model (one-time)
vibedev models download qwen-coder-1.5b

# Chat with your data
vibedev chat --query "What are my top productivity patterns?"

# Keep model loaded for fast queries
vibedev daemon start
vibedev chat --query "How can I reduce costs?"
vibedev daemon stop

    Available Models:

    • qwen-coder-0.5b - Fast, 1GB
    • qwen-coder-1.5b - Balanced, 3GB (recommended)
    • qwen-coder-3b - High quality, 6GB
    • deepseek-coder-1.3b - Efficient, 2.6GB

    GPU Acceleration:

    cargo build --release --features cuda   # NVIDIA
cargo build --release --features metal  # Apple Silicon

    Interactive TUI

    vibedev tui

    Real-time visualization inspired by dust:

    • Storage breakdown by tool
    • Session counts and token estimates
    • Cost analysis
    • Keyboard navigation (↑↓ to navigate, Enter to expand, q to quit)

    Analysis Report

    vibedev analyze --output report.md

    Sample output:

    # AI Coding Tools - Analysis Report

## Global Metrics
- Total Storage: 1.01 GB
- Total Sessions: 1,020
- Estimated Tokens: 21,245,300
- Peak Usage Hour: 14:00 UTC
- Most Used Tool: Claude Code

## Cost Estimate
- Monthly Cost: $165.71
- Optimization Potential: $49.71 (30%)

## Recommendations
1. Backup debug logs (345 MB compressible)
2. Clean file-history older than 30 days
3. Consider token caching for repeated queries

    Supported Tools

    AI Coding Assistants:

    • Claude Code, Cursor, Windsurf, Cline, Continue.dev
    • GitHub Copilot, Sourcegraph Cody, Amazon Q
    • Kiro, Roo Code, Kilo, Aider

    Code Completion:

    • Tabnine, Supermaven, CodeWhisperer, CodeGPT, Bito AI

    Logs detected from:

    • ~/.claude/ - Claude Code
    • ~/.cursor/ - Cursor
    • ~/.config/Code/User/globalStorage/ - VSCode extensions
    • ~/.config/Kiro/ - Kiro
    • And 80+ more locations

    Dataset Export

    Create sanitized datasets for fine-tuning:

    vibedev prepare --output ~/datasets

    Sanitization removes:

    • API keys (OpenAI, Anthropic, GitHub, AWS, etc.)
    • Passwords and credentials
    • Email addresses and phone numbers
    • Personal file paths
    • IP addresses

    See PREPARE_GUIDE.md for details.

    Build from Source

    git clone https://github.com/openSVM/vibedev.git
cd vibedev
cargo build --release
./target/release/vibedev --help

    Requirements:

    • Rust 1.82+
    • ~8MB binary size

    Configuration

    vibedev works out of the box with no configuration. Optional settings:

    # Filter to specific tool
vibedev analyze --tool claude

# Time range
vibedev analyze --days 30

# Output format
vibedev analyze --format json --output data.json

    Contributing

    Contributions welcome! Please read the code of conduct and submit PRs.

    # Run tests
cargo test

# Run with debug logging
vibedev --verbose analyze

    License

    MIT License - see LICENSE for details.

    Built with Rust. Inspired by the AI coding revolution.