Skip to content

juanitto-maker/AIworkbench

Repository files navigation

AIWB (AI Workbench) 🤖

Tired of one-shot AI prompts that miss the mark? Stop wrestling with single models. Start orchestrating them.

AIWB is a command-line toolkit for developers who want to elevate their AI-driven workflow. It introduces a powerful Generator-Verifier loop, turning a simple prompt into a sophisticated, multi-stage collaboration between different AI models.

License: GPL v3 Issues Last Commit


💡 The Core Concept: AI Collaboration

The magic of AIWB is its unique feedback loop. It mimics a professional developer and code reviewer team, resulting in higher-quality, refined output.

graph TD
    subgraph AIWB Workflow
        A[User Prompt] --> B{Generator};
        B -- Drafts --> C[Code/Text Draft];
        C --> D{Verifier};
        D -- Critiques --> E[Feedback & Revisions];
        E --> F{AIWB Refines Prompt};
        F -- Instructs --> B;
    end
Loading

✨ Features at a Glance

Feature Description
Mode-Based Workflows Powerful /make, /tweak, and /debug modes with structured, multi-step AI workflows for complex tasks.
Hybrid AI Engine Go beyond single-model prompting. Our unique Generator-Verifier loop uses multiple AIs to create, critique, and improve work autonomously.
Multi-Provider Support Support for Gemini, Claude, OpenAI, Groq, xAI/Grok, and Ollama (local models) - switch providers on the fly.
GitHub Integration Full GitHub support like Claude Code - clone repos, manage issues, create PRs, commit & push, all from the command line.
Smart Cost Control Built-in cost estimation and tracking. View real-time spending breakdown by provider and never get surprised by API bills.
Context Management Upload files and directories to provide context to AI. Supports multiple file types and intelligent context building.
Built for the Command Line Fast, keyboard-driven interface with beautiful TUI using gum, designed for power users on Linux, macOS, and Android (Termux).
Open & Extensible Modular architecture with plugin support. Easily add new models, templates, and extend functionality.

🚀 Get Started in 3 Steps

Get up and running in minutes.

1. Install AIWB

# One-line install (recommended)
curl -fsSL https://raw.githubusercontent.com/juanitto-maker/AIworkbench/main/install.sh | bash

# Or manual install
git clone https://github.com/juanitto-maker/AIworkbench.git
cd AIworkbench
./install.sh

# Fresh (purged) install for Termux (recommended)
curl -fsSL https://raw.githubusercontent.com/juanitto-maker/AIworkbench/main/install-termux.sh | bash
# After install completes, start using aiwb from any directory:
aiwb

Purge (Complete Removal)

# Linux/macOS: remove all AIWB files, binaries, and workspace data
rm -f ~/.local/bin/aiwb ~/.local/bin/aiwb_headless
rm -rf ~/.local/bin/lib/
rm -rf ~/.aiwb/
sed -i '/# AIworkbench installer/d' ~/.bashrc 2>/dev/null || true
sed -i '/export PATH=.*\.local\/bin/d' ~/.bashrc 2>/dev/null || true
source ~/.bashrc

# Or use the uninstall script (interactive — will prompt for confirmation)
curl -fsSL https://raw.githubusercontent.com/juanitto-maker/AIworkbench/main/uninstall.sh | bash

Purged Reinstall (Clean Slate)

# Linux/macOS: purge then reinstall in one sequence
rm -f ~/.local/bin/aiwb ~/.local/bin/aiwb_headless && \
rm -rf ~/.local/bin/lib/ ~/.aiwb/ && \
curl -fsSL https://raw.githubusercontent.com/juanitto-maker/AIworkbench/main/install.sh | bash

# Termux: the Termux installer includes a full nuclear cleanup automatically
curl -fsSL https://raw.githubusercontent.com/juanitto-maker/AIworkbench/main/install-termux.sh | bash

2. Add to PATH

Ensure ~/.local/bin is on your PATH:

export PATH="$HOME/.local/bin:$PATH"
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

3. Configure API Keys

# Start AIWB and use the interactive setup
aiwb
# Then type: /keys

# Or configure manually
aiwb keys

🤖 AI Assistant Usage

When using Claude Code or similar AI assistants with this codebase:

IMPORTANT: Read context/docs/CLAUDE_QUICK_REF.md before starting work.

This ensures efficient context loading and preserves rate limits by instructing the AI to:

  • Load only 1-3 files initially
  • Expand context incrementally as needed
  • Focus on specific files mentioned in your request
  • Ask before loading additional dependencies

Quick start: Begin your session with: Read context/docs/CLAUDE_QUICK_REF.md first, then [your task]


🗺️ Roadmap

We're on a mission to build the ultimate command-line AI assistant. Join us!

For a detailed breakdown of our future plans, see the docs/ROADMAP.md file.


💻 Quick Usage Examples

Interactive Chat

# Start interactive mode
aiwb

# Try these commands
> /make              # Enter make mode to generate code
> /quick Create a Python password generator
> /wizard            # Guided workflow for beginners
> /help              # Show all commands

Mode-Based Workflows

aiwb
> /make              # Enter make mode
make> prompt Create a REST API for user management
make> model          # Choose your AI model
make> uploads ./docs/ ./src/
make> check Focus on security and error handling
make> run            # Execute with cost estimation

Quick Commands

# One-shot generation with verification
aiwb quick "Create a sorting algorithm in Python"

# Start with wizard (great for beginners)
aiwb wizard

# Check system health
aiwb doctor

GitHub Integration (Like Claude Code)

# Interactive GitHub menu
aiwb github

# Quick commands
aiwb github status          # Show git status
aiwb github commit "msg"    # Commit changes
aiwb github push            # Push to remote
aiwb github pr create       # Create pull request
aiwb github issue create    # Create issue

# In chat mode
> /github                   # Open GitHub menu
> /github status            # Quick status check

Configuration

# Manage API keys (includes GitHub token)
aiwb keys

# Configure settings
aiwb settings

# Check costs
aiwb costs

For detailed documentation, see QUICKSTART.md, docs/GITHUB_INTEGRATION.md, and docs/.

❤️ Join the Crew & Support

This is an open-source project built by and for the community. Please consider contributing or supporting our work.

Sponsor Ko-fi

This project is licensed under the GPL-3.0 License - see the LICENSE file for details.

About

No description, website, or topics provided.

Resources

Code of conduct

Contributing

Stars

Watchers

Forks

Releases

No releases published

Sponsor this project

Packages

 
 
 

Contributors

Languages