A Python package for building intelligent multi-agent systems using CrewAI
CrewAIMaster is an advanced framework that automatically generates, manages, and executes multi-agent crews based on natural language task descriptions. It provides a CLI interface and comprehensive backend system for creating intelligent AI agents with memory, tools, and safety guardrails.
# install from PyPI (when available) pip install crewaimaster # Or Install from source (recommended for development) git clone https://github.com/VishApp/crewaimaster cd crewaimaster python -m venv venv source venv/bin/activate pip install -e .
# Install Python 3.10+ python --version # Configure your LLM provider (see supported providers) crewaimaster providers # Quick setup with OpenAI (most common) crewaimaster providers --configure openai --api-key "your-openai-key" --model "gpt-4"
# Create an intelligent crew using AI analysis crewaimaster create "Write a comprehensive market analysis report for electric vehicles in 2024" --name electric_vehicles_market_analysis_crew
# Run the crew (requires configured LLM provider) crewaimaster run electric_vehicles_market_analysis_crew # With additional context: crewaimaster run electric_vehicles_market_analysis_crew --input "Focus on Tesla, BMW, and Volkswagen specifically"
Generated crews can also be executed directly using environment variables:
# Navigate to the generated crew directory cd crews/electric_vehicles_market_analysis_crew # Run using standard environment variables export OPENAI_API_KEY="your-openai-key" ./run.sh "your input" # Or run using CrewAIMaster-specific environment variables export CREWAIMASTER_LLM_PROVIDER="openai" export CREWAIMASTER_LLM_MODEL="gpt-4" export CREWAIMASTER_LLM_API_KEY="your-openai-key" export CREWAIMASTER_LLM_BASE_URL="https://api.openai.com/v1"
flowchart LR
A["`**1. Task Definition**
Natural Language Task`"] --> B["`**2. AI Analysis**
π€ Task Complexity
π― Agent Requirements
π οΈ Tool Selection`"]
B --> C["`**3. Crew Creation**
π₯ Agent Design
π§ Tool Assignment
π Task Orchestration`"]
C --> D["`**4. Execution**
π Multi-Agent Coordination
π Real-time Processing
π Progress Monitoring`"]
D --> E["`**5. Results & Analytics**
π Output Generation
π Performance Metrics
πΎ Persistent Storage`"]
E --> F["`**6. Optimization**
π§ Crew Modification
β‘ Performance Tuning
π€ Export/Backup`"]
F --> G["`**7. Reuse & Scale**
π Crew Reusability
π Knowledge Building
π Production Deployment`"]
classDef stepStyle fill:#f9f9f9,stroke:#333,stroke-width:2px,color:#333
class A,B,C,D,E,F,G stepStyle
CrewAIMaster follows a clean, layered architecture designed for intelligent multi-agent system creation and execution:
flowchart TD
%% User Entry Point
User[π€ User Input<br/>Natural Language Task] --> CLI[π₯οΈ CLI Interface<br/>crewaimaster create/run/providers]
%% Configuration Layer
CLI --> Config[βοΈ Configuration<br/>config.yaml<br/>LLM Providers]
%% AI Orchestration Core
CLI --> MasterAgent[π§ Master Agent<br/>Intelligent Orchestrator]
%% AI Analysis Pipeline
MasterAgent --> TaskAnalyzer[π Task Analyzer<br/>β’ Complexity Assessment<br/>β’ Requirements Extraction<br/>β’ Agent Planning]
TaskAnalyzer --> AgentDesigner[π₯ Agent Designer<br/>β’ Role Definition<br/>β’ Tool Selection<br/>β’ Capability Mapping]
AgentDesigner --> CrewOrchestrator[π Crew Orchestrator<br/>β’ Team Assembly<br/>β’ Process Selection<br/>β’ Workflow Design]
%% Core Generation Engine
CrewOrchestrator --> CrewDesigner[π§ Crew Designer<br/>File-Based Generator]
Config --> CrewDesigner
CrewDesigner --> FileGen[π File Generator<br/>β’ Project Structure<br/>β’ Python Modules<br/>β’ YAML Configs]
%% Output Generation
FileGen --> GeneratedFiles{π Generated Crew Project}
%% Generated Project Structure
GeneratedFiles --> AgentYAML[agents.yaml<br/>Agent Definitions]
GeneratedFiles --> TaskYAML[tasks.yaml<br/>Task Specifications]
GeneratedFiles --> CrewPY[crew.py<br/>CrewAI Implementation]
GeneratedFiles --> MainPY[main.py<br/>Execution Entry Point]
%% Execution Runtime
MainPY --> CrewAI[π CrewAI Runtime<br/>Multi-Agent Execution]
CrewAI --> AgentA[π€ Agent A<br/>Specialized Role]
CrewAI --> AgentB[π€ Agent B<br/>Specialized Role]
CrewAI --> AgentC[π€ Agent C<br/>Specialized Role]
%% Tool Integration
AgentA --> Tools[π οΈ Tool Registry<br/>β’ Web Search<br/>β’ File Operations<br/>β’ Code Execution<br/>β’ Custom Tools]
AgentB --> Tools
AgentC --> Tools
%% LLM Integration
Config --> LLMProvider[π LLM Provider<br/>β’ OpenAI<br/>β’ Anthropic<br/>β’ Google<br/>β’ Custom APIs]
LLMProvider --> AgentA
LLMProvider --> AgentB
LLMProvider --> AgentC
LLMProvider --> MasterAgent
%% Memory & Knowledge
CrewAI --> Memory[π§ Memory System<br/>β’ Agent Memory<br/>β’ Shared Context<br/>β’ Knowledge Base]
%% Safety & Guardrails
Tools --> Guardrails[π‘οΈ Guardrails<br/>β’ Safety Checks<br/>β’ Content Filtering<br/>β’ Validation]
%% Final Output
CrewAI --> Results[π Results<br/>Task Completion<br/>Generated Content]
%% Styling
classDef userLayer fill:#e8f5e8,stroke:#1b5e20,stroke-width:3px,color:#000
classDef cliLayer fill:#e1f5fe,stroke:#01579b,stroke-width:2px,color:#000
classDef aiLayer fill:#f3e5f5,stroke:#4a148c,stroke-width:2px,color:#000
classDef coreLayer fill:#fff8e1,stroke:#ff8f00,stroke-width:2px,color:#000
classDef fileLayer fill:#fce4ec,stroke:#880e4f,stroke-width:2px,color:#000
classDef runtimeLayer fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000
classDef toolLayer fill:#f1f8e9,stroke:#33691e,stroke-width:2px,color:#000
class User userLayer
class CLI,Config cliLayer
class MasterAgent,TaskAnalyzer,AgentDesigner,CrewOrchestrator aiLayer
class CrewDesigner,FileGen,LLMProvider coreLayer
class GeneratedFiles,AgentYAML,TaskYAML,CrewPY,MainPY fileLayer
class CrewAI,AgentA,AgentB,AgentC,Memory,Results runtimeLayer
class Tools,Guardrails toolLayer
- User Input: Natural language task description via CLI
- AI Analysis: Master Agent analyzes complexity and requirements
- Intelligent Design: AI agents design optimal crew composition
- Code Generation: Automated creation of CrewAI project files
- Execution: Generated crew runs with real-time coordination
- Results: Task completion with generated content and insights
CrewAIMaster's architecture is designed for scalability, modularity, and intelligent automation:
- CLI Interface: Rich terminal experience with typer and rich libraries
- Command Processing: Handles user commands and provides interactive feedback
- Input Validation: Ensures commands are properly formatted and validated
- MasterAgentCrew: Main orchestrator using AI agents for intelligent decision-making
- TaskAnalyzerAgent: Advanced NLP analysis of user tasks and requirements
- AgentDesignerAgent: Intelligent design of agents based on task requirements
- CrewOrchestratorAgent: Optimizes crew composition and execution strategies
- CrewDesigner: Handles CrewAI integration and agent instantiation
- TaskAnalyzer: Legacy fallback for task analysis with pattern matching
- Tool Registry: Centralized management of all available tools
- Available Tools: Comprehensive library of built-in and custom tools
- Guardrails: Safety and validation systems for secure operation
- CrewAI Engine: Core execution engine for running multi-agent crews
- Agent Memory: Sophisticated memory management for agent learning and context
- User Input β CLI processes commands and validates input
- AI Analysis β MasterAgentCrew analyzes task using specialized AI agents
- Crew Creation β CrewDesigner instantiates agents with appropriate tools
- Execution β CrewAI Engine runs the crew with real-time monitoring
CrewAIMaster uses a .crewaimaster/config.yaml configuration file for all settings. Environment variables are no longer supported - all configuration must be done via CLI commands or direct config file editing.
# See all supported providers and configuration examples
crewaimaster providersConfigure any supported provider using the CLI:
OpenAI:
crewaimaster providers --configure openai --api-key "your-openai-key" --model "gpt-4" # Automatically sets base_url to https://api.openai.com/v1
Anthropic:
crewaimaster providers --configure anthropic --api-key "your-anthropic-key" --model "claude-3-sonnet-20240229" # Automatically sets base_url to https://api.anthropic.com/v1
Google:
crewaimaster providers --configure google --api-key "your-google-key" --model "gemini-pro" # Automatically sets base_url to https://generativelanguage.googleapis.com/v1beta
DeepSeek:
crewaimaster providers --configure deepseek --api-key "your-deepseek-key" --model "deepseek-chat" # Automatically sets base_url to https://api.deepseek.com/v1
Custom Provider:
crewaimaster providers --configure custom --api-key "your-key" --base-url "https://api.example.com/v1" --model "gpt-4o-mini" # Requires explicit base_url for custom endpoints
We welcome contributions! Here's how to get started:
- Fork the repository
- Create a feature branch:
git checkout -b feature/amazing-feature - Make your changes and add tests
- Run tests:
pytest tests/ - Commit changes:
git commit -m 'Add amazing feature' - Push to branch:
git push origin feature/amazing-feature - Open a Pull Request
# Clone and setup development environment git clone https://github.com/VishApp/crewaimaster cd crewaimaster # Install development dependencies pip install -e .
CrewAIMaster is released under the MIT License. See LICENSE for details.
- CrewAI - Core multi-agent framework
- LangChain - LLM integration tools
- Sentence Transformers - Text embeddings
- FAISS - Vector similarity search
Built with β€οΈ for the AI community