Skip to content

BLERBZ/blerbz-plugins

BranchesTags

Repository files navigation

Blerbz Plugins for Claude Code

Enhance your Claude Code experience with confidence scores, intelligent auto-continuation, and multi-agent planning.

License: MIT Claude Code

What Are These Plugins?

Three powerful plugins that supercharge Claude Code for developers working on any project:

🎯 Inference Confidenz

Get real-time confidence scores (0-99%) on every Claude response. Know when to trust the output and when to dig deeper.

Your response from Claude...

---
CZ 87%  🟢

🔄 Inference Continuez

Intelligent auto-continuation based on confidence thresholds. Claude proceeds automatically when confident, asks you when uncertain.

Default: Auto-continues at 70%+ confidence, asks below that threshold.

📋 Inference Planz

Transform vague prompts into actionable implementation roadmaps through a multi-agent workflow:

  1. Research Agent - Analyzes intent and identifies unknowns
  2. Survey Agent - Generates interactive clarification questions
  3. Plan Agent - Creates production-grade implementation plan

Who Is This For?

Any developer using Claude Code to build projects:

  • 🚀 Next.js / React developers
  • 🐍 Python / FastAPI / Django developers
  • 📦 Node.js / TypeScript developers
  • 🦀 Rust / Go / any language developers
  • 🎨 Full-stack, frontend, backend - works with all stacks

Quick Start

Install in Your Project (Recommended)

Navigate to your project and run:

cd /path/to/your/project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

Restart Claude Code and you're done! ✅

Global Install (All Projects)

Install once for all projects on your machine:

curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash

See QUICKSTART.md for detailed installation options.

Features

Inference Confidenz

See confidence scores on every response:

/inference-confidenz:status              # Check plugin status
/inference-confidenz:configure enable    # Enable display
/inference-confidenz:configure disable   # Disable display
/inference-confidenz:help                # Full documentation

Score breakdown:

  • 🔴 0-32%: Low confidence (proceed with caution)
  • 🟡 33-65%: Medium confidence (verify important details)
  • 🟢 66-99%: High confidence (likely accurate)

Inference Continuez

Intelligent auto-continuation:

/inference-continuez:continuez show      # Show current settings
/inference-continuez:confidence 85       # Set threshold to 85%
/inference-continuez:continuez enable    # Enable auto-continue
/inference-continuez:continuez disable   # Disable auto-continue

Customize threshold:

# Permissive (continues more often)
export INFERENCE_CONTINUEZ_THRESHOLD=70

# Conservative (asks more often)
export INFERENCE_CONTINUEZ_THRESHOLD=90

Project-specific settings:

echo '{"confidence_threshold": 85}' > .claude/continuez.config.json

Inference Planz

Multi-agent planning workflow:

/inference-planz:run Build user authentication
/inference-planz:planz <prompt>          # Alias
/inference-planz:status                  # Show configuration
/inference-planz:help                    # Full documentation

Example:

/inference-planz:run Build a REST API for blog posts with authentication

→ Research Agent identifies: auth method, database choice, API design
→ Survey Agent asks: JWT or session-based? PostgreSQL or MongoDB?
→ Plan Agent generates: Step-by-step implementation roadmap

Installation Modes

Project-Scoped (Recommended)

Install plugins directly in your project:

cd your-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

Benefits:

  • ✅ Different projects can have different plugin versions
  • ✅ Team members get plugins automatically when cloning
  • ✅ Project-specific threshold configurations
  • ✅ No global configuration conflicts

Files created:

your-project/
├── .claude/
│   ├── plugins/
│   │   └── blerbz-plugins/
│   ├── installed_plugins.json
│   ├── settings.json
│   └── *.config.json (optional)

Global (Cross-Project)

Install once for all projects:

curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash

Benefits:

  • ✅ Works across all projects immediately
  • ✅ Single installation and configuration
  • ✅ Simpler for solo developers

Files created:

~/.claude/
├── plugins/
│   ├── known_marketplaces.json
│   └── installed_plugins.json
└── settings.json

Use Cases

1. Solo Developer

Quick global install:

curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash

Now all your projects have confidence scores and auto-continue!

2. Team Project

Project-scoped for team collaboration:

cd your-team-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

# Commit to repo
git add .claude/
git commit -m "Add Claude Code plugins"
git push

Team members clone and immediately get:

  • Confidence scores
  • Auto-continuation
  • Planning workflows

3. Different Thresholds Per Project

High-stakes project (conservative):

cd ~/banking-app
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project
echo '{"confidence_threshold": 90}' > .claude/continuez.config.json

Rapid prototyping (permissive):

cd ~/hackathon-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project
echo '{"confidence_threshold": 70}' > .claude/continuez.config.json

4. Open Source Project

Add plugins to your open source repo:

cd your-open-source-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

# Add to .gitignore
cat >> .gitignore <<'EOF'

# Claude Code plugins
.claude/plugins/**/.git
.claude/*.config.json
EOF

git add .claude/ .gitignore
git commit -m "Add Claude Code plugins for contributors"
git push

Contributors get enhanced Claude experience automatically!

Examples by Tech Stack

Next.js / React

cd ~/my-nextjs-app
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

# Now ask Claude:
# "Add a new API route with error handling"
# → Get confidence scores on responses
# → Auto-continue if confidence is high
# → Use /inference-planz:run for complex features

Python / FastAPI

cd ~/my-fastapi-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

# Example usage:
/inference-planz:run Build user registration endpoint with email verification

Node.js / TypeScript

cd ~/my-typescript-api
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

Rust / Go / Any Language

cd ~/my-rust-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

Works with any language or framework - plugins enhance Claude Code itself, not specific to any tech stack.

Updating Plugins

Project-Scoped Updates

cd your-project
curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

The script auto-updates plugins when re-run.

Global Updates

cd ~/blerbz-plugins
git pull origin main

Uninstalling

Quick Uninstall

# Project-scoped
./.claude/plugins/blerbz-plugins/uninstall.sh --project

# Global
~/blerbz-plugins/uninstall.sh --global

# Auto-detect
./uninstall.sh

Uninstall Options

./uninstall.sh --help        # Show help
./uninstall.sh --dry-run     # Preview what will be removed
./uninstall.sh --force       # Skip confirmation prompts
./uninstall.sh --project     # Uninstall project-scoped only
./uninstall.sh --global      # Uninstall global only

The uninstall script will:

  1. Remove plugin entries from installed_plugins.json
  2. Remove enabled entries from settings.json
  3. Remove marketplace entry (global only)
  4. Delete the plugin directory
  5. Remove any config files (.claude/*.config.json)

Restart Claude Code after uninstalling to complete the removal

Configuration

Configuration Files

Project-scoped:

  • .claude/plugins/installed_plugins.json - Plugin registry
  • .claude/settings.json - Enabled plugins
  • .claude/continuez.config.json - Continuez threshold (optional)
  • .claude/confidenz.config.json - Confidenz settings (optional)

Global:

  • ~/.claude/plugins/known_marketplaces.json - Marketplace registry
  • ~/.claude/plugins/installed_plugins.json - Global plugins
  • ~/.claude/settings.json - Global enablement
  • ~/.config/inference-continuez/settings.json - Global threshold

Precedence: Project-scoped configs override global configs.

Adjust Auto-Continue Threshold

Via command:

/inference-continuez:confidence 85

Via environment variable:

export INFERENCE_CONTINUEZ_THRESHOLD=85

Via project config:

echo '{"confidence_threshold": 85}' > .claude/continuez.config.json

Recommendations:

  • 70%: Default (balanced, good for most development workflows)
  • 80%: Moderate (asks more often, better for production code)
  • 90%: Conservative (high-stakes projects, financial/medical systems)

Architecture

Plugin Structure

blerbz-plugins/
├── .claude-plugin/
│   └── marketplace.json          # Marketplace manifest
├── inference-confidenz/          # Confidence scores
│   ├── .claude-plugin/
│   │   └── plugin.json           # Plugin manifest
│   ├── hooks/
│   │   ├── hooks.json            # Hook definitions
│   │   └── confidence-hook.sh    # Hook executor
│   ├── skills/                   # Slash command skills
│   ├── config/                   # Default configuration
│   └── docs/                     # Documentation
├── inference-continuez/          # Auto-continuation
│   ├── .claude-plugin/
│   │   └── plugin.json
│   ├── hooks/
│   │   └── hooks.json
│   ├── scripts/
│   │   ├── confidence-evaluator.py
│   │   └── settings-manager.py
│   ├── skills/
│   └── TROUBLESHOOTING.md
├── inference-planz/              # Multi-agent planning
│   ├── .claude-plugin/
│   │   └── plugin.json
│   ├── hooks/
│   │   └── hooks.json
│   ├── scripts/
│   │   ├── planz-orchestrator.py
│   │   └── planz-summary.py
│   ├── skills/
│   ├── tests/                    # Unit & integration tests
│   └── examples/
├── quickstart.sh                 # Installation script
├── uninstall.sh                  # Uninstallation script
├── QUICKSTART.md                 # Quick start guide
├── CHANGELOG.md                  # Version history
└── README.md                     # This file

How It Works

Inference Confidenz:

  1. Intercepts Claude responses
  2. Analyzes response characteristics
  3. Calculates confidence score (0-99%)
  4. Appends score to response output

Inference Continuez:

  1. Detects stopping points in Claude's work
  2. Evaluates continuation confidence
  3. Auto-proceeds if confidence >= threshold
  4. Asks user when confidence < threshold or operation is risky

Inference Planz:

  1. Accepts vague/complex task description
  2. Research Agent analyzes intent
  3. Survey Agent generates clarification questions
  4. User answers questions
  5. Plan Agent creates detailed implementation roadmap

Troubleshooting

Plugins not loading

  1. Restart Claude Code (required after installation)
  2. Verify installation: /inference-confidenz:status
  3. Check installed_plugins.json paths are correct
  4. Check settings.json has plugins enabled

No confidence scores appearing

/inference-confidenz:status
/inference-confidenz:configure enable

Restart Claude Code.

Auto-continue not working

/inference-continuez:continuez show

Check threshold setting. Destructive operations are blocked by design.

Which installation mode am I using?

# Project-scoped
ls .claude/plugins/installed_plugins.json

# Global
ls ~/.claude/plugins/installed_plugins.json

See QUICKSTART.md for more troubleshooting tips.

Documentation

Contributing

Contributions are welcome! Please:

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/amazing-feature)
  3. Commit your changes (git commit -m 'Add amazing feature')
  4. Push to the branch (git push origin feature/amazing-feature)
  5. Open a Pull Request

License

MIT License - see LICENSE file for details.

Support

Credits

Built with ❤️ by the Blerbz team.

Special thanks to:

  • Anthropic for Claude Code
  • The Claude Code community
  • All contributors and users

Enhance your Claude Code experience today! 🚀

curl -fsSL https://raw.githubusercontent.com/Blerbz/blerbz-plugins/main/quickstart.sh | bash -s -- --project

About

Claude Code Super Powers

Resources

Readme

License

MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •  

Languages