Agent skill

internal-red-team-audit

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/internal-red-team-audit

SKILL.md

Internal Red Team Audit

Skill Metadata

  • Name: internal-red-team-audit
  • Category: Adversarial Testing
  • Version: 1.0.0

Purpose

Execute comprehensive security and alignment audits from an internal red team perspective. This methodology challenges assumptions, identifies vulnerabilities, and stress tests systems for potential failure modes.

Protocol

Phase 1: Reconnaissance

  1. Gather system context and current state
  2. Identify attack surfaces and potential weak points
  3. Document assumptions being made

Phase 2: Threat Modeling

  1. Enumerate potential threat actors and their capabilities
  2. Map attack vectors and exploitation paths
  3. Prioritize risks based on impact and likelihood

Phase 3: Adversarial Testing

  1. Execute controlled probes against identified weaknesses
  2. Document findings with evidence
  3. Classify severity (Critical/High/Medium/Low/Info)

Phase 4: Synthesis Report

  1. Summarize findings in structured format
  2. Provide remediation recommendations
  3. Generate risk score and executive summary

Output Format

json
{
  "audit_id": "UUID",
  "timestamp": "ISO_8601",
  "scope": "[AUDIT_SCOPE]",
  "findings": [
    {
      "id": "FINDING-001",
      "severity": "HIGH|MEDIUM|LOW|INFO",
      "title": "Finding Title",
      "description": "Detailed description",
      "evidence": "Supporting evidence",
      "recommendation": "Mitigation steps"
    }
  ],
  "risk_score": 0.0-10.0,
  "executive_summary": "Brief overview"
}

Behavioral Calibration

yaml
sycophancy_level: 0.0
pushback_threshold: 0.1
adversarial_mode: true
trust_but_verify: true

Example Usage

  • "Audit this API design for security vulnerabilities"
  • "Red team analysis of authentication flow"
  • "Identify failure modes in this distributed system"

Didn't find tool you were looking for?

Be as detailed as possible for better results