Flow - Intelligent Skill Orchestrator

Flow allows users to express build ideas or tasks in natural language, finds the best existing skills, scans them for security, and compiles them into a single executable FLOW skill.

Features

• Natural Language Processing: Describe what you want to build in plain English
• Skill Registry: Reusable object architecture - never reinvent the wheel
• Security Scanning: Integrated security checks for malware, spyware, and malicious code
• Smart Composition: Automatically combines skills into unified workflows
• Dependency Resolution: Handles skill dependencies with topological sorting

Installation

Prerequisites

• Python 3.8+
• pip

Clone the Repository

git clone https://github.com/bvinci1-design/flow.git
cd flow

Install Dependencies

pip install -r requirements.txt

How to Run

Web UI (Streamlit Interface)

For the most user-friendly experience, use the Streamlit web interface:

streamlit run streamlit_ui.py

This will open a web interface in your browser with:

• 🏠 Build Flow: Enter natural language requests and watch the 5-step process
• 📚 Skill Registry: Browse, search, and filter available skills
• ℹ️ About: Learn more about Flow's architecture and features

The web UI provides real-time feedback on:

• Intent parsing results
• Skill matching scores
• Security scan status (✅ Safe / ⚠️ Warning / ❌ Unsafe)
• Composition progress

In Clawdbot

1. Import the Flow skill into Clawdbot
2. Use natural language to describe what you want:

   Flow> Build a web scraper that extracts product prices and saves to CSV
   

3. Flow will:
   • Parse your request
   • Search for existing skills
   • Security scan all components
   • Compose a unified FLOW skill

On Any Device (CLI)

Interactive Mode

python flow.py

Then type your requests:

Flow> Create an automation that monitors a webpage for changes
Flow> quit

Single Command

python flow.py "Build a data pipeline that fetches API data and stores in database"

List Available Skills

python flow.py --list

Get Skill Info

python flow.py --info skill_name

Architecture

flow/
├── flow.py                      # Main orchestrator
├── natural_language_parser.py   # NLP for user intent
├── skill_registry.py            # Reusable skill database
├── skill_scanner_integration.py # Security scanning
├── skill_composer.py            # Compiles skills into FLOW
├── README.md
└── SKILL.md                     # ClawdHub skill definition

How It Works

1. Parse Request: NLP parser extracts intent, capabilities, and steps
2. Search Registry: Finds existing skills matching requirements
3. Security Scan: Checks all skills for malicious patterns
4. Compose: Merges skills into single executable FLOW
5. Register: Saves new FLOW for future reuse

Security Features

Flow includes comprehensive security scanning:

• Code execution detection (eval, exec)
• Data exfiltration patterns
• Crypto mining indicators
• System modification attempts
• Obfuscation detection
• AST-based analysis

Configuration

Create a flow_config.json:

{
  "skills_directory": "./skills",
  "output_directory": "./flows",
  "registry_path": "./skill_registry.json",
  "security_level": "standard",
  "auto_update_registry": true
}

Examples

from flow import Flow

# Create Flow instance
flow = Flow()

# Process a natural language request
result = flow.process(
    "Build a skill that monitors YouTube channels and sends notifications"
)

print(f"Created: {result.flow_name}")
print(f"Skills used: {result.skills_used}")
print(f"Security: {result.security_status}")

License

MIT

Author

@bvinci1-design