Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

SENTINEL-1 ULTIMATE (COMPLETE)

v2.0.0

Portable USB AI agent with 7 guardians for offline sovereignty on Windows

0· 100·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for kofna3369/sentinel-1-ultimate.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "SENTINEL-1 ULTIMATE (COMPLETE)" (kofna3369/sentinel-1-ultimate) from ClawHub.
Skill page: https://clawhub.ai/kofna3369/sentinel-1-ultimate
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install sentinel-1-ultimate

ClawHub CLI

Package manager switcher

npx clawhub@latest install sentinel-1-ultimate
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
high confidence
!
Purpose & Capability
The SKILL.md advertises a complete portable USB package (koboldcpp executable, agent Python files, launcher, vault folder, MODELS_INSTALL.md) yet the registry manifest contains only SKILL.md. A skill that truly provides a portable offline agent would legitimately include those files or declare concrete download/install steps; this one does neither. That mismatch is disproportionate to the stated purpose.
!
Instruction Scope
The document gives only high-level user-facing steps (copy folder, run Lancer_Sentinel.bat, download models) and marketing claims ("zero trace", "BitLocker encryption", "7 Guardians") but contains no concrete runtime instructions for the AI agent and no safe, verifiable download sources. It implicitly directs the user to fetch unspecified large binaries/models (MODELS_INSTALL.md referenced but not present), which is a risky open-ended instruction.
!
Install Mechanism
There is no install spec and no code files in the package, yet the description claims bundled executables and model binaries. That forces users to obtain binaries from unspecified external locations if they follow the README—downloading executables from unknown sources is high risk. The lack of declared, reputable install hosts or checksums is a red flag.
!
Credentials
The skill requests no environment variables or credentials, which is consistent with an instruction-only package — but it claims features (BitLocker vault, 'zero trace', environment verification, memory validation) that typically require elevated privileges or system-level access on Windows. The skill neither requests nor documents how those privileges would be obtained or justified, creating a mismatch.
!
Persistence & Privilege
The package asserts system-level behaviors (encryption, guardians, anonymous mode) that imply persistent or privileged actions on the host, yet the skill metadata shows no persistent files and 'always' is false. The claimed ability to run 'zero trace' from USB or manipulate BitLocker suggests elevated privilege requirements that are not reflected in the manifest or instructions.
What to consider before installing
Do not run or trust unknown .bat or executable files referenced in the marketing text. Before installing or following the README, ask the publisher for: (1) a complete file manifest and cryptographic checksums, (2) the exact, reputable URLs and signatures for any model or binary downloads, and (3) proof of authorship or a homepage/source repository. If you want to try it, do so only in an isolated VM or a disposable test machine; never enable BitLocker or run unfamiliar installers on a machine holding important data. Prefer packages that include binaries in the registry bundle or that point to well-known release pages (signed GitHub releases, vendor sites) with verifiable checksums. Because this package is missing the files it claims to include and gives vague privileged instructions, treat it as untrusted until the above details are provided.

Like a lobster shell, security has layers — review code before you run it.

betavk971t6aqd9v65czefgn7fww191855mz4latestvk971t6aqd9v65czefgn7fww191855mz4
100downloads
0stars
2versions
Updated 1w ago
v2.0.0
MIT-0

SENTINEL-1 ULTIMATE

STATUS: BETA
VERSION: Complete Package (includes koboldcpp)
PLATFORM: Windows 10/11 (USB 3.1)

What is SENTINEL-1?

A portable AI agent that runs from a USB drive with zero installation, zero trace, and full offline capability. Includes 7 security guardians for maximum protection.

Features

  • 7 Guardians (Morgana's security checks)
  • BitLocker encryption on vault folder
  • Offline mode — no internet required once models installed
  • Portable — runs on any Windows PC
  • Sovereign — your data stays with you
  • Includes Koboldcpp (622MB, ready to run)

What's Included

  • koboldcpp/ — Koboldcpp executable (622MB) ✅
  • agent/sentinel/ — Sentinel agent Python files
  • skills/ — Skills system
  • Lancer_Sentinel.bat — Windows launcher
  • README_EN.md — Complete English documentation
  • MODELS_INSTALL.md — Model download instructions
  • docs/ — Quick guide + Update ritual
  • vault/ — Encrypted storage

Quick Start

  1. Copy SENTINEL-1-EN folder to USB drive
  2. Run Lancer_Sentinel.bat
  3. Download models (see MODELS_INSTALL.md)
  4. Start chatting!

Security

7 Guardians verify every launch:

  1. BitLocker Vault Protection
  2. Pre-Chasse Test (Gemma validation)
  3. USB Speed Check
  4. Environment Verification
  5. Memory Validation
  6. Anonymous Mode
  7. Guardian Protocol

Credits

Axioma Cluster (Ezekiel build, Morgana verification, Merlin guardian) For Alexandre's brother — sovereign AI in your pocket.

Built with ❤️ by Axioma Stellaris - 2026

Comments

Loading comments...