iflow-mcp_cycloarcane-hostile-command-suite 0.1.3

Intelligent OSINT investigation system with MCP tools

Hostile Command Suite - OSINT Package

Author: cycloarcane
Contact: [e-mail]
License: PolyForm Noncommercial License 1.0.0

Intelligent Open Source Intelligence Investigation System

A terminal-based OSINT investigation framework with AI-powered analysis and intelligent agent decision-making. Features automated profile scraping, multi-platform username investigation, and local LLM integration for enhanced intelligence gathering.

---

🚀 Quick Start

Prerequisites

1. Install ollama (for AI analysis):

curl -fsSL https://ollama.ai/install.sh | sh
ollama pull qwen3:8b  # recommended model

2. Install OSINT tools:

# Arch Linux
yay -S sherlock-git mosint

# Ubuntu/Debian  
pip install sherlock-project
# For mosint, download from: https://github.com/alpkeskin/mosint

Installation

git clone https://github.com/cycloarcane/Hostile-Command-Suite.git
cd Hostile-Command-Suite
python -m venv .venv
source .venv/bin/activate  # Linux/Mac
pip install -r requirements.txt

Usage

Interactive comprehensive investigation:

python3 HCSO.py --interactive
# Then provide ALL target information: names, usernames, emails, addresses, etc.

Command line investigation:

# Single targets (backward compatibility)
python3 HCSO.py cycloarcane
python3 HCSO.py [e-mail]

# Comprehensive targets (multiple data points)
python3 HCSO.py "John Smith, @johnsmith123, [e-mail], works at Acme Corp"
python3 HCSO.py --model llama3.2 "Jane Doe [e-mail] https://linkedin.com/in/janedoe"

---

🛠️ Features

Core Capabilities

• 🔍 Username Investigation: Sherlock integration across 400+ social media platforms
• 📧 Email Investigation: Mosint integration for email intelligence and breach analysis
• 🌐 Profile Scraping: Automated extraction of profile details from discovered accounts
• 🔍 Web Search Intelligence: DuckDuckGo search integration for comprehensive OSINT gathering
• 🤖 AI Agent: Local ollama integration for intelligent decision-making and analysis
• ⚡ Intelligent Workflow: Automatic tool chaining and investigation pivoting
• 🎨 Rich Terminal: Professional red/black themed interface with progress indicators

Supported Targets

Target Type	Primary Tool	Secondary Tools	AI Analysis
Username	Sherlock → Profile Scraper	DuckDuckGo Search, Link Analyzer	✅ Full Analysis
Email	Mosint	DuckDuckGo Search	✅ Full Analysis
Any Target	DuckDuckGo Search	Context-dependent pivoting	✅ Full Analysis

MCP Tool Architecture

The system uses Model Context Protocol (MCP) based tool servers:

• sherlock_server.py: Username investigation across platforms
• mosint_server.py: Email enumeration and breach investigation
• profile_scraper_server.py: Intelligent profile content extraction
• duckduckgo_server.py: Web search for comprehensive intelligence gathering
• link_analyzer_server.py: Deep analysis of URLs and GitHub profiles

AI-Powered Intelligence

The AI agent provides:

• Comprehensive Data Extraction: Uses LLM to parse and categorize all provided target information
• Intelligent Tool Selection: Automatically chooses appropriate tools based on data types:
  • Names → DuckDuckGo web search for public records and news
  • Usernames → Sherlock for social media platform discovery
  • Emails → Mosint for breach data and domain analysis
  • Organizations → Web search for corporate intelligence
  • URLs → Link analyzer for deep content analysis
• Investigation Pivoting: Discovers new leads and suggests follow-up actions
• Security Risk Assessment: Evaluates exposure levels and security implications
• Pattern Recognition: Identifies connections across platforms and data sources
• Decision Making: Determines when investigations are complete vs need continuation

---

🏗️ Architecture

Intelligent Agent Design

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   HCSO Agent    │──▶│  Ollama AI       │───▶│ Investigation   │
│                 │    │  Decision Engine │    │ Recommendations │
└─────────────────┘    └──────────────────┘    └─────────────────┘
         │
         ▼
┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│ MCP Tool        │──▶│  Tool Results    │───▶│ Profile Scraper │
│ Manager         │    │  Analysis        │    │ Auto-Trigger    │
└─────────────────┘    └──────────────────┘    └─────────────────┘
         │
         ▼
┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│ Sherlock        │    │ Mosint           │    │ Profile         │
│ Username Search │    │ Email Intel      │    │ Scraper         │
└─────────────────┘    └──────────────────┘    └─────────────────┘

Investigation Workflow

1. Comprehensive Input: User provides ALL available target information (names, usernames, emails, addresses, organizations, URLs, etc.)
2. AI Data Extraction: LLM parses and categorizes information into structured data types
3. Intelligent Tool Selection: System automatically selects appropriate tools for each data type:
   • Names → Web search for public intelligence
   • Usernames → Social media platform discovery
   • Emails → Breach analysis and domain intelligence
   • Organizations → Corporate and public records search
   • URLs → Deep content and profile analysis
4. Parallel Investigation: Multiple tools execute simultaneously based on extracted data
5. AI Analysis: Intelligent analysis of all findings and cross-reference discovery
6. Decision Point: AI recommends additional investigations or marks complete
7. Iterative Enhancement: Follow-up investigations based on discovered leads

---

📊 Example Output

  ╔═════════════════════════════════════════════════════════════════════════╗
  ║  ██   ██ ███████ ██████         ███████ ███████  ██ ███    ██ ████████  ║
  ║  ██   ██ ██      ██             ██   ██ ██       ██ ████   ██    ██     ║
  ║  ███████ ███████ ██      █████  ██   ██ ███████  ██ ██ ██  ██    ██     ║
  ║  ██   ██      ██ ██             ██   ██      ██  ██ ██  ██ ██    ██     ║
  ║  ██   ██ ███████ ██████         ███████ ███████  ██ ██   ████    ██     ║
  ╚═════════════════════════════════════════════════════════════════════════╝

      Hostile Command Suite - OSINT Package
      Intelligent Open Source Intelligence Investigation System

  Using AI Model: qwen3:8b
  Available Tools: sherlock, mosint, profile_scraper, duckduckgo_search, link_analyzer

  ──────────────────────────────────────────────────────────────────────────────

  ═══ COMPREHENSIVE TARGET INFORMATION ═══
  Provide ALL available information about your target for intelligent analysis
  Include: names, usernames, emails, addresses, organizations, social profiles, etc.

  Enter ALL target information: John Smith, @johnsmith123, [e-mail], works at TechCorp

  Analyzing Provided Information
  ──────────────────────────────────────────────────────────────────────────────

   Extracted Target Intelligence 
  ┏━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
  ┃ Data Type      ┃ Extracted Values                                                  ┃
  ┡━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
  │ Names          │ John Smith                                                        │
  │ Usernames      │ johnsmith123                                                      │
  │ Emails         │ [e-mail]                                           │
  │ Organizations  │ TechCorp                                                          │
  └────────────────┴───────────────────────────────────────────────────────────────────┘
  ────────────────────────────────────────────────────────────────

  Investigating Name: John Smith

   SHERLOCK Investigation Results 
  ┏━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┓
  ┃ Metric         ┃ Value       ┃
  ┡━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━┩
  │ Target         │ cycloarcane │
  │ Accounts Found │ 17          │
  │ Status         │ Success     │
  └────────────────┴─────────────┘
  ────────────────────────────────────────────────────────────

  Found 17 profiles, scraping for additional intelligence...
  PROFILE_SCRAPER Investigation Results
  ┏━━━━━━━━━━━━━━━━━━┳━━━━━━━━━┓
  ┃ Metric           ┃ Value   ┃
  ┡━━━━━━━━━━━━━━━━━━╇━━━━━━━━━┩
  │ Total Scraped    │ 5       │
  │ Successful       │ 4       │
  │ With Useful Info │ 4       │
  │ Status           │ Success │
  └──────────────────┴─────────┘
  ────────────────────────────────────────────────────────────

  AI Agent Analyzing...
  ╭─ AI Investigation Analysis ──────────────────────────────────────────────╮
  │ ANALYSIS: Investigation revealed GitHub profile with security research    │
  │ interests (LLMs + red team). High-value intelligence gathered from        │
  │ multiple platforms. Profile scraping provided sufficient context.        │
  │                                                                           │
  │ RECOMMENDATION: Investigation complete - sufficient intelligence gathered │
  │ TOOL: NONE                                                                │
  │ TARGET: N/A                                                               │
  │ REASONING: Profile analysis reveals technical expertise and security      │
  │ focus. No additional tools needed for current investigation scope.        │
  ╰───────────────────────────────────────────────────────────────────────────╯

---

🛡️ Security & Ethics

Responsible Use

• Legal Compliance: All investigations must comply with applicable laws
• Authorization: Only investigate targets you have permission to research
• Rate Limiting: Respects platform rate limits and implements delays
• Local Processing: All AI analysis happens locally via ollama (no cloud)

Privacy & Security

• No Data Persistence: Investigation results are not stored long-term
• Local LLM: AI analysis never leaves your machine
• Tool Isolation: Each OSINT tool runs independently
• Professional Focus: Designed for defensive security and legitimate research

---

🔧 Configuration

AI Model Selection

# Use different ollama models
python3 HCSO.py --model llama3.2 target
python3 HCSO.py --model qwen3:8b target
python3 HCSO.py --model mixtral target

Configurable Prompts

Agent behavior is configurable via YAML files:

• prompts/agent_system.yaml: Core agent instructions and tool selection logic
• prompts/tool_prompts.yaml: Tool-specific analysis templates

Tool Capabilities

Tool	Input	Capabilities	Auto-Trigger
DuckDuckGo Search	Names, Organizations	Web intelligence, news, public records	Auto for names
Sherlock	Username	400+ platform search	Auto for usernames
Mosint	Email	Breach data, domain intel	Auto for emails
Link Analyzer	URLs	GitHub profiles, web content analysis	Auto for URLs
Profile Scraper	URLs	Bio, followers, verification	After Sherlock

---

🚧 Development

Adding New OSINT Tools

1. Create MCP server in mcp_tools/new_tool_server.py
2. Add tool detection in MCPToolManager.check_available_tools()
3. Implement tool calling in MCPToolManager.call_tool()
4. Add result display in display_investigation_result()
5. Update agent prompts for tool selection logic

Architecture Benefits

Previous Complex Architecture:

• 15+ microservices with FastMCP
• PostgreSQL database requirement
• Web UI and API complexity
• Multiple authentication layers

Current Intelligent Agent Architecture:

• Single intelligent agent with MCP tools
• No database required
• Pure terminal interface with AI
• Local ollama for decision-making
• Automatic tool chaining and pivoting

---

📈 Roadmap

Phase 1: Core Intelligence ✅

• Sherlock username investigation with AI analysis
• Mosint email investigation with AI analysis
• Intelligent profile scraping from social media
• AI-powered investigation decision making
• MCP-based tool architecture

Phase 2: Enhanced Analysis 🚧

• Link analyzer for deep GitHub/social media analysis
• Domain investigation capabilities
• Phone number OSINT integration
• Correlation analysis across findings

Phase 3: Advanced Intelligence 📋

• Investigation session management
• Custom tool integration framework
• Automated investigation workflows
• Advanced AI reasoning and pivoting

---

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/new-tool)
3. Add your MCP tool server following existing patterns
4. Update agent prompts for tool integration
5. Test with various target types
6. Submit a pull request

Development Setup

git clone https://github.com/cycloarcane/Hostile-Command-Suite.git
cd Hostile-Command-Suite
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python3 HCSO.py --interactive

---

📄 License

This project is licensed under the PolyForm Noncommercial License 1.0.0 - see the LICENSE file for details.

Copyright © cycloarcane ([e-mail])

---

Intelligent. Terminal. Effective.

Advanced OSINT investigation with AI-powered decision making

For questions or feature requests, contact [e-mail]