benchjack

Official

Audit AI benchmarks for hackability.

Authorbenchjack
Version1.0.0
Installs0

System Documentation

What problem does it solve?

BenchJack provides a security-audit workflow for AI benchmarks to uncover vulnerabilities in evaluation pipelines, such as data leakage, isolation issues, and prompt-injection risks.

Core Features & Use Cases

  • Automated security audit of AI benchmarks to reveal evaluation vulnerabilities (V1–V8).
  • Static + AI hybrid analysis using a suite of checks and custom rules to map risk across tasks.
  • Generates structured findings and task mappings to guide authors in hardening benchmarks and improving reliability.

Quick Start

Run the BenchJack scanner on a benchmark repository to reveal evaluation vulnerabilities.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: benchjack
Download link: https://github.com/benchjack/benchjack/archive/main.zip#benchjack

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.