Hashcat MCP Server

An AI-powered password hash auditing assistant that combines a local LLM with Hashcat through the Model Context Protocol (MCP).

Built for authorized security audits — the LLM autonomously identifies hash types, selects attack strategies, escalates through multiple approaches, and reports results in plain English. All processing is local — no data leaves your machine.

---

Architecture

User (natural language)
        ↓
Local LLM — qwen2.5 via Ollama
  (decides which tools to call and with what arguments)
        ↓
MCP Client — bridges LLM decisions to real tool execution
        ↓
MCP Server — exposes Hashcat as structured tools
        ↓
Hashcat — performs actual password recovery
        ↓
Results logged to results/session_log.json

---

Tools

Tool	Description
identify_hash	Identifies hash algorithm from format/length (MD5, NTLM, SHA-1, bcrypt etc.)
list_wordlists	Lists available wordlists and rule files on the system
run_hashcat	Cracks a single hash with configurable attack mode and wordlist
crack_batch	Cracks multiple hashes from a file with automatic escalation

Attack Escalation (automatic)

The LLM follows this sequence for every hash automatically:

1. Dictionary attack — 10k-common wordlist
2. Rule-based attack — 100k-ncsc + best66.rule
3. Hybrid attack — 100k-ncsc + 4-digit mask (?d?d?d?d)
4. Larger wordlist — 000webhost

---

Stack

Component	Version	Purpose
Hashcat	v7.1.2	Password recovery engine
Ollama	latest	Local LLM runtime
qwen2.5	7B	Tool-calling capable LLM
FastMCP (Python)	latest	MCP server framework
WSL2 Ubuntu	24.x	Linux environment on Windows

---

Setup

Prerequisites

• Windows with WSL2 + Ubuntu
• Ollama installed (curl -fsSL https://ollama.com/install.sh | sh)
• Hashcat installed (sudo apt install hashcat)
• SecLists cloned (git clone --depth 1 https://github.com/danielmiessler/SecLists.git ~/SecLists)

Install

git clone https://github.com/YOUR_USERNAME/hashcat-mcp.git
cd hashcat-mcp
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Pull the LLM

ollama pull qwen2.5:latest

---

Usage

Start the assistant

cd hashcat-mcp
source venv/bin/activate
python3 mcp_client.py

Crack a single hash

You: crack this hash: 482c811da5d5b4bc6d497ffa98491e38
Assistant: [calls identify_hash → MD5 identified]
           [calls run_hashcat → password123 found]
           The password is: password123

Crack a batch file

Create a file with one hash per line, or username:hash format:

admin:e90664c0af74160644d29e4d6147969b
user1:5f4dcc3b5aa765d61d8327deb882cf99
482c811da5d5b4bc6d497ffa98491e38

Then ask:

You: crack all hashes in /home/user/hashes.txt — they are MD5

Output:

==================================================
BATCH CRACK SUMMARY
==================================================
Total hashes:   3
Cracked:        3 (100.0%)
Not cracked:    0
==================================================
RESULTS:
  ✓ admin      → Summer2024
  ✓ user1      → password
  ✓ (hash)     → password123
==================================================
Full results saved to: results/session_log.json

---

Session Logging

Every crack attempt is automatically logged to results/session_log.json:

{
  "timestamp": "2026-08-19T19:33:27.932857",
  "username": "admin",
  "hash": "e90664c0af74160644d29e4d6147969b",
  "hash_mode": 0,
  "cracked": true,
  "password": "Summer2024",
  "attack_used": "mode 6 wordlist=100k-ncsc mask=?d?d?d?d"
}

---

Supported Hash Types

Algorithm	Hashcat Mode	Example Format
MD5	-m 0	482c811da5d5b4bc6d497ffa98491e38
NTLM	-m 1000	cc36cf7aa9dc7f2d2c2c43f877c32b0c
SHA-1	-m 100	cbfdac6008f9cab4083784cbd1874f76618d2a97
SHA-256	-m 1400	ef92b779... (64 chars)
bcrypt	-m 3200	$2a$10$...
sha512crypt	-m 1800	$6$...

---

Security & Ethics

This tool is designed exclusively for authorized password security audits:

• Testing organizational password strength against real-world attack methods
• Identifying weak passwords before attackers do
• Generating evidence for security policy enforcement

Never use this tool on hashes you do not have explicit authorization to audit.

---

Project Structure

hashcat-mcp/
├── mcp_server.py        # MCP server — exposes Hashcat as tools
├── mcp_client.py        # MCP client — connects LLM to server
├── prompts/
│   └── system_prompt.txt  # LLM behavior instructions
├── hashes/              # Hash files for testing
├── results/
│   └── session_log.json   # Automatic audit log
└── requirements.txt