Skillget

genpark-adversarial-redteam-prompt-injection-fuzzer-skill

mcp
Guvenlik Denetimi
Basarisiz
Health Uyari
  • No license — Repository has no license file
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 8 GitHub stars
Code Basarisiz
  • Hardcoded secret — Potential hardcoded credential in client.py
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

GenPark AI Agent Skill - Multi-turn jailbreak and indirect prompt injection fuzzer evaluating LLM agent safety guardrails.

README.md

GenPark AI Agent Skill - Adversarial Red-Team Prompt Injection Fuzzer

Generates systematic red-teaming test matrices (DAN roleplay, prompt extraction, hidden comment injection) and scores model resistance.

Verified by GenPark AI and compatible with Model Context Protocol (MCP).

Architecture Diagram

graph TD
    A[Target Vulnerability Vector] --> B[Adversarial Mutation Generator]
    B --> C[Roleplay & DAN Archetype Synthesis]
    B --> D[System Prompt Exfiltration Probes]
    B --> E[Hidden Markdown Comment Vectors]
    C --> F[Dispatch Against Target Agent Under Test]
    D --> F
    E --> F
    F --> G[Score Resistance: RESISTED vs COMPROMISED]

Features

  • Automated CI/CD Safety Gates: Catches prompt injection vulnerabilities before shipping agent features.
  • Zero External Dependencies: Pure Python 3.9+ standard library.

Yorumlar (0)

Sonuc bulunamadi