Verfy V1

Developed by Rowhanissick & Verfy
Base Architecture: empero-ai/Qwythos-9B-v2 (Qwen3.5 9B Hybrid-Attention)
Portfolio: rowhanissick.onrender.com

Overview

Verfy AI v1 is an autonomous OS-agent and reasoning model engineered specifically for autonomous desktop/server computer-use, security auditing, and system operations.

Unlike standard conversational chatbots that only output unstructured text, Verfy natively generates structured execution blocks directly in its token stream. This allows agent daemons to reliably parse internal reasoning, step planning, safety checks, and environment actions (mouse, keyboard, CLI) without brittle regex workarounds.

Native Block Architecture

Verfy AI v1 communicates and executes tasks using structured XML-style tags:

Block Tag	Purpose	Example
`<think>`	Deep internal reasoning & strategy planning	Analyzing UI state and formulating goals
`<steps>`	Decomposes operations into sequential tasks with `<title>`	`1. Locate browser 2. Dispatch click`
`<checks>`	Precondition validation & safety checks with `<title>`	`- Screen bounds verified<br>- Non-destructive check passed`
`<screenshot>`	Requests visual frame capture of screen/window	`capture(mode='screen', region='desktop')`
`<command>`	Shell / Bash terminal command execution	`netstat -tulnp \| grep 8080`
`<read>`	Safe file, log, or memory dump inspection	`read_file(path='/var/log/syslog', lines=50)`
`<write>`	File patching and exploit/payload deployment	`write_file(path='/tmp/patch.py', content='...')`
`<input>`	GUI mouse movements and keystroke dispatches	`mouse_move(x=500, y=500)` `mouse_click('left')`
`<vuln>`	Flags identified CVEs & security vulnerabilities	`ID: VULN-001 Severity: HIGH CWE-78`

Headline Capabilities

1,048,576 Token Context: Full 1M-token context window out of the box via YaRN rope-scaling, allowing whole-codebase and long agentic trajectory tracking.
Uncensored Technical Reasoning: Built on top of Qwythos-9B-v2, Verfy does not refuse cybersecurity research, penetration testing, reverse engineering, or system auditing tasks.
Dual Interaction Modes: Operates both at the CLI level (<command>) and the GUI level (<input> + <screenshot>).

How to Use (Python)

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

MODEL_ID = "Rowhanissick/Verfy-9B"

tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    torch_dtype=torch.float16,
    device_map="auto",
    trust_remote_code=True
)

prompt = "Find the terminal on desktop, open it, and check the network interfaces."

messages = [
    {"role": "user", "content": prompt}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.6,
    top_p=0.9
)

response = tokenizer.decode(outputs[0][len(inputs.input_ids[0]):], skip_special_tokens=True)
print(response)