⚓ Navy AI

An AI-powered persistent CLI agent with built-in pentest, recon, and system tools.
Supports Ollama · LM Studio · OpenAI · Gemini · Anthropic — switch models and routes on the fly.

---

Features

• Interactive & argument mode — chat in a session or fire a one-liner from your terminal
• Multi-provider — Ollama, LM Studio, OpenAI, Gemini, Anthropic — all in one tool
• Local AI first — Ollama and LM Studio supported out of the box with configurable server routes
• Reasoning engine — the agent generates a step-by-step plan before executing multi-step tasks and tracks progress visually
• 3-phase agent loop — Gather → Act → Verify architecture keeps the agent structured and on-task
• Smart query routing — simple factual questions answered instantly without tool overhead
• Arrow-key approval — confirm or decline tool calls with ← → keys, no typing required
• Built-in pentest tools — port scanner, SSL checker, HTTP prober, subdomain enum, WHOIS, and more
• MCP-based tool server — extensible, runs as a local subprocess
• Session management — save, load, and export conversations as Markdown
• Audit log — every command and response is logged locally
• Loop & dead-command detection — stops runaway tool loops; remembers which native tools are unavailable and switches to WSL automatically

---

Install

pip install navy-ai

Install with your preferred AI provider:

pip install "navy-ai[ollama]"      # local models via Ollama
pip install "navy-ai[openai]"      # GPT-4o, o3, o4-mini — also required for LM Studio
pip install "navy-ai[gemini]"      # Gemini 1.5 / 2.0
pip install "navy-ai[anthropic]"   # Claude Sonnet / Opus / Haiku
pip install "navy-ai[all]"         # every provider at once

---

Usage

Interactive mode

navy

⚓ ~ ❯ what ports are open on 10.0.0.1?
⚓ ~ ❯ summarise the files in this folder
⚓ ~ ❯ am i running the latest kernel?

Argument mode (single-shot)

navy "what is the name of this computer"
navy "how long has this machine been running"
navy --model gpt-4o "scan ports on 10.0.0.1"
navy --yes "what processes are using the most CPU"

Options

Flag	Description
--model <name|alias>	Override the model (name or alias from models.json)
--ctx <int>	Context window size (default: 32768)
--yes / -y	Skip all confirmation prompts

In-session commands

Command	Description
model <alias>	Switch model mid-session
/models	List all model aliases
/route [url]	Show or change the local AI server URL (Ollama or LM Studio)
/route ollama <url>	Change the Ollama server URL
/route lms <url>	Change the LM Studio server URL
continue	Give the agent +10 more turns to finish a long task
/save [name]	Save the current session
/load <name>	Load a saved session
/sessions	List saved sessions
/export [file]	Export transcript as Markdown
/reset	Clear conversation memory
exit / quit	Exit Navy

---

Configuration

On first run Navy auto-creates ~/.config/navy/models.json. Edit it to set your default model and API keys.

models.json

{
  "default": "qwen2.5:14b",
  "providers": {
    "openai":    { "api_key": "" },
    "gemini":    { "api_key": "" },
    "anthropic": { "api_key": "" },
    "ollama":    { "host": "http://127.0.0.1:11434" },
    "lmstudio":  { "host": "http://localhost:1234/v1" }
  },
  "presets": {
    "kimi":    "kimi-k2.5:cloud",
    "gpt4o":   "gpt-4o",
    "flash":   "gemini-2.0-flash",
    "sonnet":  "claude-sonnet-4-5",
    "qwen14":  "qwen2.5:14b",
    "lms":     "lms:local-model"
  }
}

API keys can also be set via environment variables:

export OPENAI_API_KEY=sk-...
export GEMINI_API_KEY=AIza...
export ANTHROPIC_API_KEY=sk-ant-...

config.json — timeouts and tool limits

{
  "server": {
    "command_timeout": 120,
    "max_command_timeout": 1800
  },
  "cli": {
    "max_turns": 15,
    "max_response_tokens": 4096
  }
}

Tip: For long-running commands like nmap -p-, the AI automatically requests a longer timeout. The cap is max_command_timeout (default 30 min).

---

Local AI

Ollama

Install Ollama, pull a model, and run navy:

ollama pull qwen2.5:14b
navy

Change the server address if Ollama runs on a remote machine:

/route ollama http://192.168.1.10:11434

LM Studio

1. Download LM Studio and load any GGUF model
2. Open the Local Server tab and click Start Server
3. Install the openai package: pip install "navy-ai[openai]"
4. Use lms: prefix to select the model:

navy --model lms:meta-llama-3.1-8b-instruct

Or switch mid-session:

model lms:qwen2.5-7b-instruct

Change the LM Studio server address:

/route lms http://192.168.1.10:1234/v1

The model name after lms: must match what LM Studio reports — check GET http://localhost:1234/v1/models or the LM Studio UI.

---

Built-in Tools

Tool	Description
execute_command	Run shell commands with per-command timeout control
read_file / write_file	Local file read/write
search_files	Find files by name or content
fetch_url	Fetch and parse web pages
search_web	DuckDuckGo search
scan_ports	TCP port scanner (supports ranges: 1-65535)
http_probe	HTTP status + response headers
check_security_headers	Security header grading (A–F)
ssl_check	TLS certificate validity, expiry, ciphers
dns_lookup	DNS resolution
whois_lookup	Domain registrar, expiry, nameservers
subdomain_scan	DNS-based subdomain enumeration
get_system_specs	GPU / RAM / CPU information
get_security_logs	Windows Security & Defender events

---

Pentest Workflow

Navy follows a structured recon workflow when given a target:

scan_ports → http_probe → ssl_check → subdomain_scan → whois_lookup

Important: Only use pentest tools against targets you are authorized to test.

---

License

MIT © Zrnge