attacker

Attack observable product behavior to find reproducible breakages missed by test suites.

5|Updated Jun 5, 2026
One-click install
npx skills add https://github.com/VincentJiang06/skills --skill attacker-vincentjiang06
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: attacker
Source: https://github.com/VincentJiang06/skills/tree/main/skills/attacker
Command: npx skills add https://github.com/VincentJiang06/skills --skill attacker-vincentjiang06

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill eliminates the blind spot created by correlated errors and false-positive test suites, where a green TDD suite sits on top of a broken product or an idea contains undetected flaws. It provides engineered independence to find what the builder's own context misses.

Core Features & Use Cases

  • Adversarial product attacks: Mount scoped attacks on a feature's observable behavior (not its source) to find reproducible breakages.
  • Idea red-teaming: Red-team arguments, designs, or plans by attacking the steelmanned claim from an independent perspective.
  • Fresh-context independence: Runs as a new subagent that never sees the implementation, tests, or author framing, ensuring the defect found is real.
  • Use Case: A team shipping a new checkout flow wants to know what their tests missed. Use this skill to attack the observable behavior and receive a handoff document set of proven breaks for the fix round.

Quick Start

Use the attacker skill to red-team the checkout feature from a fresh context and hand me the proven, reproducible attack records.

Frequently Asked Questions about attacker

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I red-team a product feature to find bugs my test suite misses?

To red-team a product feature, you mount scoped attacks on its observable behavior from a fresh context that excludes source code and tests, yielding reproducible breakage records for fixing.

What is adversarial testing for software features and how does it work?

Adversarial testing for software features applies an independent perspective to uncover correlated errors that pass standard tests. It works by attacking observable behavior using a fresh subagent without implementation source access.

Can I red-team an abstract design or plan to find flawed assumptions?

Yes, you can red-team an abstract design or plan by attacking the steelmanned claim from an independent perspective, uncovering undetected flaws and flawed assumptions that the original author's context missed.

Why does my TDD suite pass but the product is still broken?

A TDD suite passes while the product is broken due to correlated errors creating a blind spot. Adversarial product attacks provide engineered independence to find what the builder's own context misses.

Do I need to provide source code for an adversarial attack on my product?

No, you do not provide source code for an adversarial attack. The skill requires a fresh subagent context that explicitly excludes implementation source and test suites to ensure valid, reproducible breakages.

How do I validate reproducible breakages found during red-teaming?

You validate reproducible breakages found during red-teaming by enforcing a scope contract and emitting machine-checkable attack records with deterministic validation, ensuring the defects found are real.