OFFSITE.DARK
← Greps

Jun 1, 2026

2 min

Uploaded by automajicly

HF Model
automajicly/Local_Security_Model
Base Model
Qwen/Qwen2.5-1.5B-Instruct-GGUF
Pipeline
text-generation
Downloads
0
Likes
1
  • pentest
  • autonomous-agent
  • mcp
  • kali
  • flask
  • qwen2.5

pentest

Community upload on Hugging Face — indexed for inquiry, not endorsement.

Local Security Model — Autonomous Pentest Agent Framework

Uploaded to Hugging Face by automajicly — Qwen2.5 1.5B-based project tagged as local pentest agent with MCP and Kali Linux integration.

What Is This?

Uploaded to Hugging Face by automajicly. Local_Security_Model is tagged as a full project artifact — not just weights — combining a fine-tuned Qwen2.5 1.5B model with Python/Flask infrastructure for local autonomous security testing.

Metadata

FieldValue
Authorautomajicly
Base modelQwen/Qwen2.5-1.5B-Instruct-GGUF
Libraryother (full project)
LicenseMIT
TagsMCP, Kali Linux, bug bounty, ethical hacking

Tagged Components

Model card tags suggest bundled capabilities:

  • MCP server / agent loop — Model Context Protocol tool orchestration
  • Kali Linux — offensive tooling environment integration
  • Flask — local web UI or API wrapper
  • Pentesting tools — external binary invocation

Why It Might Matter

This upload represents the agent-as-repo pattern — Hugging Face hosting not just tensors but runnable attack-surface automation. Research value:

  • Audit MCP tool permissions and command injection boundaries
  • Measure 1.5B model adequacy for multi-step pentest planning vs. larger agents (RavenX, ArmurAI)
  • Study how "ethical hacking" framing affects model refusal rates

Caution

Projects bundling agents + tool execution carry higher operational risk than static weights. Treat repository code as untrusted; run only in isolated VMs.

Attribution

Open-source model indexed for security research inquiry. Uploaded to Hugging Face by automajicly.