🧠⚡ SQLMAP SKYNET - Autonomous AI v1.2.0

AI-assisted SQLMap orchestration + real-time dashboard + MCP tool server (agent-ready)

A guided, repeatable workflow: Targets → Preflight → Phased Scan → AI Assist → Evidence → Report

---

---

🔥 Update v1.2.0 — Autonomous AI + MCPFast Commands

✅ Renamed branding from Intelligence → Autonomous AI
✅ Added MCPFast (FastMCP) command-line usage (No UI required)
✅ MCP supports URL scan and targetlist (.txt) scan
✅ Best for Windows + Linux servers (headless / no GUI)

What is SQLMAP SKYNET (simple)

SQLMAP SKYNET wraps sqlmap.py with a Neural Ops workflow — fast, structured, and trackable.

• ✅ Operational phases (Detect → Bypass → Enumerate → Dump)
• ✅ Dashboard visibility (logs, stats, artifacts)
• ✅ Autonomous AI tuning (Ollama / Cloud)
• ✅ RAG intelligence (learns from success, recalls patterns)
• ✅ MCP tool server (agent automation)

Goal: higher signal, less noise, and reports real proof.

---

workflow : Dashboard + MCP + AI + RAG

flowchart TD
  A["1. Targets<br/>single URL or list"] --> B["2. Preflight<br/>GET or POST + headers + cookies + safe limits"]

  B --> C["3A. Dashboard Control<br/>FastAPI UI + WebSocket"]
  B --> D["3B. MCP Control<br/>Agent tool calls"]

  C --> E["4. Runner Core<br/>SQLMapRunner -> SQLMapBackend -> sqlmap.py"]
  D --> E

  E --> F["5A. Autonomous AI<br/>Ollama first, cloud fallback"]
  E --> G["5B. RAG Memory<br/>learn + recall"]
  E --> H["5C. Web Intel<br/>optional search"]
  E --> I["5D. WAF Intel<br/>fingerprint + bypass hints"]

  F --> J["6. Artifacts<br/>reports + logs + sessions"]
  G --> J
  H --> J
  I --> J

  J --> K["7. View or Export<br/>Dashboard + files"]

Loading

Neural Ops Stack (Simple)

flowchart TD
  subgraph INPUT["INPUT"]
    T["Targets"] --> PF["Preflight"]
  end

  subgraph CONTROL["CONTROL"]
    UI["Dashboard"] --- MCP["MCP Client"]
  end

  subgraph ENGINE["ENGINE"]
    RUN["Runner Core"] --> SM["sqlmap.py"]
  end

  subgraph INTEL["INTEL"]
    AI["Autonomous AI"] --- RAG["RAG Memory"] --- WEB["Web Intel"] --- WAF["WAF Intel"]
  end

  subgraph OUTPUT["OUTPUT"]
    ART["Artifacts"] --> REP["Reports"]
  end

  PF --> CONTROL
  CONTROL --> ENGINE
  ENGINE <--> INTEL
  ENGINE --> OUTPUT

Loading

Key idea: MCP and Dashboard are two ways to drive the same engine logic:

• Dashboard = human operator UI
• MCP = agent/tool interface

---

📸 Screenshots of Feature

🧾 Report Samples

🎥 Demo Video

Windows requirements (choose your AI mode)

✅ Minimum (no local AI)

• Windows 10/11 64-bit
• Python 3.12+
• Git
• sqlmap.py present (recommended: .\sqlmap\sqlmap.py)
• Any CPU / 8GB RAM

Use Cloud AI (OpenAI / Groq / DeepSeek / Kimi / Claude) or run with AI features OFF.

✅ Recommended (local AI with Ollama)

• Windows 10/11 64-bit
• Python 3.12+
• 16GB RAM (recommended)
• GPU optional (faster with Nvidia GPU + VRAM)
• Ollama + Llama 3.2 model (llama3.2:latest)

Tip: Llama 3.2 in Ollama has small variants (1B/3B). If your PC is weak, local AI still works — just expect slower responses.

---

Table of contents

• Install & setup (Tested on Windows)
• Install sqlmap.py
• Install Ollama + Llama 3.2
• Run SKYNET dashboard
• First scan tutorial
• Enable MCP server (agent mode)
• Cloud AI setup (if your PC is weak)
• Config power tips (config.py)
• Full project structure map
• Artifacts
• Community & Resources
• 🤝 Contributing
• 📜 License
• ⚠️ Disclaimer

---

Install & setup (Tested on Windows)

1) Clone project

Open PowerShell:

git clone https://github.com/drcrypterdotru/sqlmap-skynet
cd sqlmap-skynet

2) Create venv (required)

py -3.12 -m venv .venv
.\.venv\Scripts\Activate.ps1
python --version

3) Install Python dependencies

pip install -U fastapi uvicorn python-dotenv aiohttp requests numpy scikit-learn

Optional (for MCP server over HTTP):

pip install -U fastmcp

4) Create .env

Copy-Item .env.example .env

Open .env and set what you need:

• Local AI: Ollama (recommended)
• Or Cloud AI keys (if PC is weak)

---

Install sqlmap.py

Best practice (inside project folder):

git clone https://github.com/sqlmapproject/sqlmap.git sqlmap

✅ You must have:

sqlmap-skynet\
  sqlmap\sqlmap.py
  main.py
  config.py
  ...

---

Install Ollama + Llama 3.2

1) Install Ollama (Windows)

• Download & install Ollama for Windows from the official site.
• After install, open a new PowerShell and verify:

ollama --version

2) Pull the model used by SKYNET (default)

Your config.py sets:

• OLLAMA_MODELS['default'] = 'llama3.2:latest'

Pull it:

ollama pull llama3.2:latest

Test:

ollama run llama3.2:latest

If you want to switch model later, edit config.py under OLLAMA_MODELS.

---

Run SKYNET dashboard

Start the server:

python main.py --host 0.0.0.0 --port 1337 --debug

Open:

• Dashboard: http://127.0.0.1:1337

---

First scan tutorial

Use a target you own / have permission to test.

Step 1 — Add targets

In the dashboard:

• Paste one URL or multiple URLs (one per line)

Step 2 — Set method

• GET for query params
• POST for login/forms

Step 3 — Add session context (optional)

• Headers (User-Agent, Authorization)
• Cookies (session token)
• POST body (for POST scans)

Step 4 — Start

Choose safe defaults:

• max_cycles = 30
• RAG = ON (recommended)
• Web Search = OFF (start simple)

Then click Start.

Step 5 — Review

After the scan:

• Check sqlmap_reports\
• Check memory\sessions\ (if RAG enabled)

---

Enable MCP server (agent mode)

What MCP does

MCP allows an AI agent to call SKYNET tools:

• sqlmap_scan → starts scan (uses sqlmap.py)
• get_scan_status → polling status/results
• get_ai_providers → shows available AI backends

1) Install MCP dependency (once)

pip install -U fastmcp

2) Start MCP server over HTTP

Run in a new PowerShell (keep dashboard in another terminal if you want both):

fastmcp run .\mcp\server.py:mcp --transport http --host 127.0.0.1 --port 8055

Your MCP endpoint:

• http://127.0.0.1:8055/mcp

3) Confirm MCP tools

fastmcp list http://127.0.0.1:8055/mcp

Does MCP work with Autonomous AI?

✅ Yes. MCP tool sqlmap_scan creates SQLMapRunner("sqlmap.py"), and the runner uses core/autonomous_ai.py.
So MCP calls automatically use your Autonomous AI + RAG logic (no module changes needed).

---

MCPFast (FastMCP) CLI Mode — No UI (Headless / Server-Friendly)

Sometimes you don’t want the dashboard (example: Linux server / VPS / headless box).
You can run SKYNET fully from command line using FastMCP — perfect when you don’t have a GUI.

📸 MCPFast Demo Screenshots

1) Start MCP server (HTTP)

Open terminal in project root:

fastmcp run .\mcp\server.py:mcp --transport http --host 127.0.0.1 --port 8055

MCP endpoint:

• http://127.0.0.1:8055/mcp

2) Verify MCP is working (Tools + Providers + Status)

Run these commands (this is the fastest proof MCP is healthy):

fastmcp list http://127.0.0.1:8055/mcp --auth none

fastmcp call http://127.0.0.1:8055/mcp get_ai_providers --auth none
fastmcp call http://127.0.0.1:8055/mcp get_scan_status --auth none

What you should see

• Tools (3): sqlmap_scan, get_scan_status, get_ai_providers
• Providers: ollama (recommended)
• Status: running=false until you start a scan

3) Run scan by URL (one target)

fastmcp call http://127.0.0.1:8055/mcp sqlmap_scan --auth none ^
  url="http://testphp.vulnweb.com/artists.php?artist=1" ^
  method="GET" ^
  max_cycles=10 ^
  ai_provider="ollama"

4) Run scan by target list file (many targets)

Create target.txt in project root (one URL per line), then:

fastmcp call http://127.0.0.1:8055/mcp sqlmap_scan --auth none ^
  targetlist="target.txt" ^
  method="GET" ^
  max_cycles=10 ^
  ai_provider="ollama"

Tip: targetlist should be inside the project folder for safety.

Cloud AI setup (if your PC is weak)

If you can't run Ollama fast (low RAM/CPU), you can use Cloud AI.

Quick setup

1. Open .env
2. Add ONE key (example OpenAI):

OPENAI_API_KEY=YOUR_KEY_HERE

Other supported keys:

• GROQ_API_KEY=...
• DEEPSEEK_API_KEY=...
• KIMI_API_KEY=...
• ANTHROPIC_API_KEY=...

SKYNET will detect which providers are available and use them when needed.

Provider priority is defined in config.py as AI_PRIORITY.

---

Config deep guide (config.py)

config.py is the brains + intel database of SKYNET.
If you want better results, this is the first file to tune.

Tip: On GitHub you can click the file: config.py → then use the browser search (Ctrl+F) for the section names below.

Quick map (what you can tune)

• ✅ AI models (Ollama model names)
• ✅ AI provider priority (Ollama first or cloud first)
• ✅ WAF bypass packs (delay/threads/tamper presets)
• ✅ Tamper script database (profiles by aggressiveness)
• ✅ Technique priority (faster signal, less noise)
• ✅ Keyword intelligence (High-Value Columns (Keyword Intel): Customize the keyword list to match your environment (examples: crypto, email, password, hash, token, apikey, secret). SKYNET will prioritize and highlight matching columns so you can identify impact faster.)

⚠️ For educational/authorized testing only. Use only on systems you own or where you have written permission.

---

Click for full details (config.py)

1) OLLAMA_MODELS — local model selection

SKYNET reads the local Ollama model name from:

OLLAMA_MODELS = {"default": "llama3.2:latest"}

What to do:

• Keep llama3.2:latest as default if it works well on your PC.
• If your PC is slow, use a smaller model (if you have it in Ollama) and set it here.

Why it matters: model choice affects speed + quality of AI plans.

2) AI_PRIORITY — which AI provider is used first

Example:

AI_PRIORITY = ["ollama","deepseek","kimi","groq","openai","claude"]

What to do:

• Want offline-first? keep ollama first.
• Want cloud-first? move your cloud provider to the front.

Why it matters: the first available provider in this list is used most often.

3) REAL_WORLD_BYPASSES — WAF/Block presets (very important)

This section is like a “bypass playbook”.
It stores presets for common conditions like:

• Cloudflare / Akamai / Imperva / Sucuri
• ModSecurity / AWS WAF
• Generic 403 / rate limiting / captcha pages

Typical preset contains:

• delay and timing
• threads (lower threads = less blocks)
• tamper_scripts (safe → aggressive)

How SKYNET uses it:

• if it detects blocking patterns (403, suspicious headers, WAF signatures), it can propose one of these presets automatically.

Operator tip:

• Start with low/no tamper + low threads.
• Increase only if you are authorized and you confirm WAF behavior.

4) TAMPER_SCRIPTS — tamper profiles (basic → aggressive)

Tamper scripts can help evade filters but they can also:

• slow scans
• increase false positives
• increase detection/noise

SKYNET organizes them into groups like:

• basic
• moderate
• aggressive
• waf_specific

Best practice:

• Use basic first.
• Use moderate only when blocked.
• Use aggressive only with permission + clear block evidence.

5) TECHNIQUE_PRIORITY — faster signal with less noise

This controls which SQLi techniques SKYNET tries to focus on first (depending on your logic).
Typical techniques:

• Error-based
• Union-based
• Boolean-based
• Time-based

Why it matters:
On many real targets, hammering time-based from the start creates slow scans + blocks.
A good priority order can produce faster confirmation and cleaner results.

6) HIGH_VALUE_COLUMNS — keyword intel that improves result quality

This is one of the strongest “operator experience” features.

HIGH_VALUE_COLUMNS contains keywords like:

• password, pass, hash
• email, phone
• token, apikey, secret
• session, cookie
• card, cc, billing
• admin, role, permission

How SKYNET uses it:

• When enumeration finds many columns, SKYNET highlights likely high-value fields.
• This makes results easier to read and reduces time wasted on irrelevant tables.

Customize it for your environment Add your own keywords, for example:

• HR: employee_id, salary, department
• Finance: invoice, receipt, balance
• E-commerce: order_id, shipment, cart
• Your apps: telegram, line_id, customer_uid

Big win: better keyword intel = faster identification of “real impact” in a report.

7) Other useful knobs you may have in config.py

Depending on your version, you may also see:

• default delays/threads limits
• retry policy / backoff
• request headers templates
• detection patterns / error signature lists
• file browser root restrictions

If you’re unsure: keep defaults and tune slowly.

---

“More Advanced Feature and Config” (recommended way)

If you want users to click from README:

• Add a GitHub link in your repo to config.py (example):
  • ./config.py
• In GitHub, it becomes clickable automatically.

Example you can paste anywhere in README:

See: [config.py](./config.py)

Full project structure map

sqlmap-skynet/
├─ 🚀 main.py
│  └─ FastAPI dashboard server + WebSocket control + scan orchestration
├─ 🧠 config.py
│  └─ AI models + provider priority + WAF bypass packs + tamper DB + keyword intel
├─ 🧩 api.py
│  └─ (optional/legacy) stats endpoint helper (older Flask-based helper)
├─ 🧪 .env / .env.example
│
├─ 🛠️ scanners/
│  ├─ 🧭 runner.py
│  │  └─ Phase engine (detect → bypass → enumerate → dump) + session state
│  └─ 🧱 sqlmap_backend.py
│     └─ Builds and executes real `sqlmap.py` commands
│
├─ 🧠 core/
│  ├─ 🤖 autonomous_ai.py
│  │  └─ AI planner + provider health + JSON plan parsing
│  ├─ 🧬 rag_memory.py
│  │  └─ Store/recall session intelligence (learn from past scans)
│  ├─ 🧾 report_generator.py
│  │  └─ Build artifacts/reports from results
│  ├─ 📡 state_manager.py
│  │  └─ Shared runtime state (running/progress/results)
│  ├─ 🛡️ waf_intel.py
│  │  └─ WAF fingerprinting + mutation ideas + learning profile
│  └─ 🧿 debug_logger.py
│     └─ Colored logs + structured tags
│
├─ 🔎 search/
│  └─ 🌐 web_search.py
│     └─ Optional web intel lookup + caching
│
├─ 🔌 mcp/
│  └─ 🛰️ server.py
│     └─ MCP tool server (sqlmap_scan, get_scan_status, get_ai_providers)
│
├─ 🧰 utils/
│  ├─ 🗂️ file_browser.py
│  │  └─ Safe file browsing inside project root (UI file picker)
│  └─ 🧩 parsers.py
│
├─ 🎛️ templates/
│  └─ 🖥️ dashboard.html
├─ 🎨 static/
│  └─ ⚙️ js/stats.js
│
├─ 🧠 memory/
│  ├─ 🧾 sessions/
│  │  └─ RAG memory snapshots (per scan)
│  ├─ 🗃️ search_cache/
│  │  └─ cached web intel queries
│  └─ 🧠 *_patterns.json
│     └─ learned patterns + bypass intelligence
│
└─ 📦 sqlmap_reports/
   └─ generated scan reports (html/json/txt depending on run)

---

Artifacts

Where your outputs go:

• sqlmap_reports\ → scan reports (proof + export)
• memory\sessions\ → RAG snapshots per session
• memory\search_cache\ → cached web intel results
• memory\*_patterns.json → learned patterns and bypass info

---

Community & Resources

Join Telegram: https://t.me/burnwpcommunity

Website: https://drcrypter.net
More tools, resources, and updates are shared on the website + community.

---

🤝 Contributing

We welcome contributions! Feel free to fork this repository, make enhancements, and open pull requests. Please check the issues page for ongoing tasks or bug reports.

---

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

---

⚠️ Disclaimer

This tool is for educational purposes only. 🏫 The creator and contributors are not responsible for any misuse or damages caused. Use responsibly, and only on systems you own or have permission for. ✅