ralph-ultra
由 dorukardahan 提供
Coding Agents & IDEs简介
Deep-dive security audit with 1,000 iterations (~4-8 hours)
描述
该技能的功能
A 1,000-iteration security audit that runs approximately 4-8 hours across 8 structured phases. Each iteration performs one targeted check with a CVSS score, confidence level, and specific remediation. Four expert personas rotate through phases covering OWASP Top 10, authentication, infrastructure, supply chain, and compliance.
Structures 1,000 individual checks into resumable phases with checkpoint saves every 50 iterations, so audits survive context limits without losing progress.
何时使用
- Pre-release security audit before shipping a major version
- Compliance audit preparation for GDPR or SOC 2 reviews
- Post-incident investigation of a compromised codebase
- Deep follow-up after a lighter audit flags suspicious patterns
- CI/CD pipeline and container security hardening review
示例工作流
User asks: Run a full deep security audit before the v2.0 release
- 1
Auto-detects stack, git state, Dockerfiles, and CI/CD config in iteration 1
- 2
Runs 100 reconnaissance checks mapping endpoints, auth flows, exposed ports, and undeclared services
- 3
Executes OWASP Top 10 deep dive across iterations 101-250 using a Code Auditor persona
- 4
Audits secrets, JWT configuration, OAuth flows, and rate limiting in phase 3, then container and network security in phase 4
- 5
Saves checkpoint to .ralph-report.md every 50 iterations and generates a scored final report at iteration 1000
.ralph-report.md with CVSS-scored findings organized by severity, proof-of-concept exploits where applicable, and specific file-level remediation steps
安装
clawhub install dorukardahan/ralph-ultra运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。