MAJESTIC-AI

# Majestic AI v1.0 🔴 ### AI-Powered Cybersecurity Automation Platform ``` ╔══════════════════════════════════════════════════════╗ ║ MAJESTIC AI v1.0 | MCP Platform | Port 8888 ║ ╚══════════════════════════════════════════════════════╝ ``` --- ## Architecture ``` AI Agent (Claude/GPT/Copilot) │ MCP Protocol ▼ mcp_client.py ← FastMCP wrapper (50+ tool definitions) │ HTTP/REST ▼ server.py ← Flask API (0.0.0.0:8888) │ ├── agents.py (12 AI agent classes) ├── cache_engine.py (LRU cache + process registry) ├── error_engine.py (10 error types + retry + degradation) ├── visual_engine.py (blood-red ANSI terminal UI) └── browser_agent.py (Selenium headless Chrome) ``` --- ## Quick Start ### 1. Install dependencies ```bash pip install -r requirements.txt ``` ### 2. Start the API server ```bash # Default (port 8888) python server.py # Custom port python server.py --port 9000 # Debug mode python server.py --debug # Via env var set HEXSTRIKE_PORT=9000 && python server.py ``` ### 3. Connect an AI agent via MCP ```bash python mcp_client.py --server http://localhost:8888 ``` ### 4. Configure Claude Desktop Copy `claude_mcp_config.json` contents into: `%APPDATA%\Claude\claude_desktop_config.json` --- ## API Endpoints (50+) | Category | Endpoints | |---|---| | Core | `GET /health`, `GET /`, `GET /banner` | | Execution | `POST /execute`, `POST /execute/stream` | | Processes | `GET /processes`, `POST /processes/{pid}/pause|resume|terminate` | | Intelligence | `POST /intelligence/analyze-target`, `select-tools`, `optimize-parameters`, `create-attack-chain`, `smart-scan`, `technology-detection`, `correlate-vulnerabilities` | | CVE | `GET /cve/lookup/{tech}`, `GET /cve/enrich/{cve_id}` | | Exploit | `POST /exploit/generate`, `GET /exploit/templates` | | Bug Bounty | `POST /bugbounty/start`, `phase-commands`, `full-recon` | | CTF | `POST /ctf/solve`, `ctf/full-solve`, `GET /ctf/categories` | | Browser | `POST /browser/navigate|screenshot|extract-dom|cookies|security-headers|full-analysis` | | Tools | `POST /tools/{tool_name}` (55+ tools) | | Files | `POST /files/write`, `GET /files/read/{name}`, `GET /files/list`, `DELETE /files/{name}` | | Visual | `POST /visual/progress|severity-card|table` | | Telemetry | `GET /cache/stats`, `GET /telemetry/performance|errors|failures` | | System | `GET /system/info`, `GET /errors` | --- ## 12 AI Agent Classes | # | Agent | Purpose | |---|---|---| | 1 | `IntelligentDecisionEngine` | Ranks tools, builds attack chains with success probabilities | | 2 | `BugBountyAgent` | Orchestrates full recon workflows | | 3 | `CTFAgent` | CTF category solver with tool flows | | 4 | `CVEIntelligenceManager` | CVE lookup and enrichment | | 5 | `AIExploitGenerator` | PoC templates (8 vuln types) | | 6 | `VulnerabilityCorrelator` | Chains vulns, calculates risk score | | 7 | `TechnologyDetector` | Fingerprints tech from headers/body | | 8 | `RateLimitDetector` | Detects 429s and rate-limit headers | | 9 | `FailureRecoverySystem` | Tracks failures, suggests fallbacks | | 10 | `PerformanceMonitor` | Avg duration, success rate per tool | | 11 | `ParameterOptimizer` | Stealth/aggressive parameter tuning | | 12 | `GracefulDegradation` | nmap→masscan→rustscan fallback chains | --- ## Bug Bounty Phases | Phase | Tools | Timeout | |---|---|---| | Subdomain Discovery | amass, subfinder, assetfinder | 300s | | HTTP Probing | httpx, nuclei | 180s | | Content Discovery | katana, gau, waybackurls, dirsearch | 600s | | Parameter Discovery | paramspider, arjun, x8 | 240s | **Vulnerability Priority:** RCE(10) → SQLi(9) → SSRF/IDOR(8) → XSS/LFI(7) → XXE(6) → CSRF(5) --- ## CTF Categories `web` | `crypto` | `pwn` | `forensics` | `reversing` | `misc` | `osint` | `stego` **PWN flow:** `checksec → gdb-peda → ropgadget → pwntools` --- ## Error Handling 10 error types with exponential backoff (immediate → 5s → 10s, 3 retries): `TIMEOUT` | `PERMISSION_DENIED` | `NETWORK_UNREACHABLE` | `RATE_LIMITED` | `TOOL_NOT_FOUND` | `INVALID_PARAMETERS` | `RESOURCE_EXHAUSTED` | `AUTHENTICATION_FAILED` | `TARGET_UNREACHABLE` | `PARSING_ERROR` --- ## Files | File | Description | |---|---| | `server.py` | Flask REST API, 50+ endpoints, 0.0.0.0:8888 | | `mcp_client.py` | FastMCP wrapper for AI agents | | `agents.py` | 12 AI agent classes | | `cache_engine.py` | LRU cache (1000 entries, 3600s TTL) | | `error_engine.py` | Error taxonomy + retry + degradation | | `visual_engine.py` | Blood-red ANSI terminal engine | | `browser_agent.py` | Selenium headless Chrome agent | | `claude_mcp_config.json` | Claude Desktop MCP config (300s timeout) | | `requirements.txt` | Python dependencies | --- > ⚠️ **Legal Notice:** For authorized security testing, bug bounty programs, and CTF competitions only.