omnimancer-cli 0.1.4

A unified command-line interface for multiple AI language models

Omnimancer CLI

Just-In-Time Engineering, Inc.

A unified command-line interface for multiple AI providers - chat with Claude, OpenAI, Gemini, and 10+ other AI models through a single, intuitive tool.

Quick Start

Installation

Using pipx (recommended):

pipx install omnimancer-cli

Using pip:

pip install omnimancer-cli

Available Commands

After installation, use any of these commands:

• omnimancer - Full command name
• omn - Quick alias ⚡
• omniman - Alternative alias

First Run

omn  # or omnimancer, or omniman

On first run, you'll be guided through setup:

🚀 Omnimancer Setup Wizard

Select a provider to configure:
1. Claude (Anthropic)
2. OpenAI  
3. Google Gemini
4. Perplexity AI
5. Ollama (Local)
...

Choose [1]: 1

Enter your Claude API key: sk-ant-...
✅ Configuration complete!

>>> Hello! How can you help me today?
🤖 Claude: I'm Claude, an AI assistant created by Anthropic...

Basic Usage

# Start Omnimancer
omn

# Start chatting
>>> What's the weather like?

# Switch models mid-conversation  
>>> /switch openai gpt-4o
>>> Now using GPT-4. How are you different?

# Check available providers and models
>>> /providers
>>> /models

# Save conversations
>>> /save my-chat

# Load previous conversations
>>> /load my-chat

# Get help
>>> /help

Agent Mode & File Operations

Omnimancer includes advanced agent capabilities that allow AI models to perform file operations with your explicit approval:

🤖 Autonomous Agent Features

• File Creation: Create new files with AI-generated content
• File Modification: Edit existing files with intelligent changes
• Code Refactoring: Restructure and improve existing code
• Documentation Generation: Create comprehensive documentation
• Project Setup: Initialize new projects with proper structure

🔒 Secure Approval System

Every file operation requires your explicit approval with:

🔍 File Operation Approval Required
📄 Creating: data_analyzer.py
📊 Risk Level: Low | 🟢 
📏 Size: 1,247 bytes (45 lines)

[Y] Approve  [N] Deny  [D] View Details  [Q] Quit

🎨 Rich Visual Interface

• Syntax Highlighting: Code displayed with proper formatting
• Diff Views: See exactly what changes before approval
• Risk Assessment: Operations rated Low/Medium/High/Critical
• Batch Operations: Handle multiple files efficiently

⚡ Quick Examples

# Ask AI to create files
>>> Create a Python script to analyze CSV data
🔍 Shows preview → [Y] to approve → ✅ File created

# Request code modifications  
>>> Add error handling to this function
🔍 Shows diff view → [Y] to approve → ✅ File updated

# Batch project setup
>>> Set up a Flask web application
🔍 Shows 8 files → [A] approve all → ✅ Project ready

📖 Full Documentation | 🛡️ Security Guide

Supported Providers

Provider	API Key Required	Best For
Claude	Anthropic Console	Complex reasoning, analysis
Claude Code	Anthropic API key	IDE integration, coding
OpenAI	OpenAI Platform	General purpose, coding
Gemini	Google AI Studio	Large context, research
Perplexity	Perplexity	Real-time web search
xAI (Grok)	xAI Console	Creative tasks, real-time info
Mistral	Mistral Platform	Code generation, efficiency
AWS Bedrock	AWS Console	AWS integration
Ollama	No API key (local)	Privacy, offline use
Azure OpenAI	Azure setup required	Enterprise
Vertex AI	Google Cloud setup	Enterprise
OpenRouter	OpenRouter	100+ models access
Cohere	Cohere Platform	Multilingual, embeddings

Commands

Core Commands

Command	Description
/help	Show all commands
/setup	Run interactive setup wizard
/quit	Exit Omnimancer
/clear	Clear screen

Model & Provider Management

Command	Description
/models	List available models
/providers	Show configured providers
/switch [provider] [model]	Change provider/model
/validate [provider]	Validate provider configurations
/health [provider]	Check provider health status
/repair [provider]	Repair provider issues
/diagnose [provider]	Run diagnostic tests

Conversation Management

Command	Description
/save [name]	Save current conversation
/load [name]	Load saved conversation
/list	List saved conversations
/history	Conversation history management

Agent & File Operations

Command	Description
/agent	Enable/disable agent mode
/agents	Manage agent configurations
/approvals	View/manage file operation approvals
/permissions	Configure security permissions

Tool Integration

Command	Description
/tools	Show available tools
/mcp	MCP server management

Model Management

Command	Description
/add-model	Add custom model
/remove-model	Remove custom model
/list-custom-models	List custom models

System

Command	Description
/status	Show system status

Configuration

Omnimancer stores encrypted configuration in ~/.omnimancer/config.json.

Manual Configuration

# Add a provider
omnimancer --config

# Or edit configuration interactively
>>> /config

Environment Variables

export ANTHROPIC_API_KEY="your-key"
export OPENAI_API_KEY="your-key"
export GOOGLE_API_KEY="your-key"
# ... then run omnimancer

Local AI with Ollama

For privacy and offline use:

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama server
ollama serve

# Download a model
ollama pull llama3.1

# Configure Omnimancer
omn
>>> /switch ollama llama3.1

Tool Integration (MCP)

Enable AI tool calling for file operations, web search, and more:

# Install UV for MCP servers
curl -LsSf https://astral.sh/uv/install.sh | sh

# Check tool status
>>> /tools
>>> /mcp status

Popular MCP tools:

• Filesystem: File operations
• Web Search: Real-time search
• Git: Repository management

Examples

Basic Chat

>>> Explain quantum computing in simple terms
🤖 Claude: Quantum computing is like having a super-powered calculator...

>>> /switch openai gpt-4o  
>>> How would you explain it differently?
🤖 GPT-4: I'd compare quantum computing to exploring a maze...

Code Generation

>>> Write a Python function to calculate fibonacci numbers
🤖 Claude: Here's an efficient implementation using memoization:

```python
def fibonacci(n, memo={}):
    if n in memo:
        return memo[n]
    if n <= 1:
        return n
    memo[n] = fibonacci(n-1, memo) + fibonacci(n-2, memo)
    return memo[n]

Model Comparison

>>> /switch claude claude-3-5-sonnet
>>> What's 15 * 24?
🤖 Claude: 15 × 24 = 360

>>> /switch openai gpt-4o
>>> What's 15 * 24?  
🤖 GPT-4: 15 × 24 = 360

Advanced Features

• Conversation Management: Save/load chat history
• Model Switching: Compare responses between providers
• Tool Calling: AI can execute code, search web, manage files
• Health Monitoring: Provider status and diagnostics
• Configuration Templates: Pre-configured setups for different use cases

Development

git clone https://gitlab.com/jite-ai/omnimancer
cd omnimancer
pip install -e ".[dev]"
pytest

Troubleshooting

Common Issues

"No providers configured"

omn  # Run setup wizard
>>> /setup

"Invalid API key"

• Check key format (Claude: sk-ant-, OpenAI: sk-, etc.)
• Verify key at provider's website
• Use /validate command to test configuration

"Ollama connection failed"

ollama serve  # Start Ollama server
ollama pull llama3.1  # Download a model

Check system health:

omn
>>> /health  # Check all providers
>>> /diagnose  # Run diagnostics
>>> /validate  # Validate configurations

Debug mode:

export OMNIMANCER_DEBUG=1
omn

License

MIT License - see LICENSE file.

Links

• [GitHub Repository]https://gitlab.com/jite-ai/omnimancer)
• Issues
• Documentation

---

Omnimancer CLI - One tool, multiple AI providers, endless possibilities.