Skill UI

This comprehensive guide details advanced techniques for circumventing Content Security Policies (CSP). It covers exploiting weaknesses in critical directives (e.g., base-uri, object-src), abusing nonces, leveraging trusted CDN endpoints, and executing code despite strict controls like script-src 'self' or 'strict-dynamic'. Essential knowledge for deep security auditing and penetration testing.

This comprehensive playbook details advanced techniques for detecting Prototype Pollution vulnerabilities in JavaScript environments (Node.js and browser). It covers various attack vectors, including using `__proto__` and `constructor.prototype` paths, and provides black-box testing payloads for JSON parsing, query string merging, and deep object assignment. Use this when analyzing applications for potential Remote Code Execution (RCE) or unintended property leakage via object merging sinks.

This advanced playbook covers escalating prototype pollution into Remote Code Execution (RCE) and Cross-Site Scripting (XSS). It details exploitation techniques across various environments, including Node.js child_process, EJS, Pug, Handlebars, and client-side libraries like jQuery and Lodash. It is designed for security researchers and penetration testers to systematic locate and exploit deep merge sinks and framework gadgets.

This playbook provides expert techniques for reverse engineering custom virtual machines and proprietary bytecode. It guides users through identifying dispatchers (switch/table), mapping unknown opcodes, reconstructing custom ISAs, and writing dedicated disassemblers. Essential for solving complex CTF challenges, analyzing protected software, or tackling custom runtime environments.

MockHunter is an advanced skill designed to audit live web pages, providing a comprehensive reality check on visible data. It analyzes every element to determine its true source: whether it is REAL (from the database), MOCK (simulated data), LLM-generated, HARDCODED, BROKEN (broken endpoints), or UNKNOWN. Ideal for auditing AI-generated UIs or reviewing proof-of-concept MVPs before customer sign-off.

This CLI alias functions as a command for gracefully canceling an active PUA loop, mirroring the functionality of the Claude Code `/pua:cancel-loop` command. It cleans up loop state and worktree references, ensuring the cancellation event is recorded. It also safely manages configuration files, preserving unknown fields in `~/.pua/config.json`.

This utility serves as a Command Line Interface (CLI) alias for managing 'flavor' profiles within AI code environments (like Claude Code). It allows users to set or select specific configuration flavors in the ~/.pua/config.json file. Crucially, it is designed to safely update the configuration by preserving any unknown fields, ensuring no unrelated settings are accidentally overwritten.

This utility serves as a command-line alias for invoking the Claude Code `/pua:kpi` function. It loads necessary configurations (pua-pro) to generate comprehensive Key Performance Indicator (KPI) or performance report cards. It manages configuration files by updating `~/.pua/config.json` while preserving unknown fields and ensuring the directory structure exists.

A structured methodology for validating non-trivial decisions by subjecting them to an adversarial, fresh-context review. This process is crucial when correctness is paramount—such as in production, security-sensitive logic, or complex architecture—by forcing a skeptical examination that aims to disprove assumptions rather than simply approve them.

A comprehensive pre-production audit tool that scans a codebase for critical issues across eight dimensions: security, database, deployment, code quality, AI/LLM usage, dependencies, frontend, and observability. It intercepts deployment commands, forcing users to resolve critical failures before shipping code. The audit is stack-agnostic and provides detailed pass/fail/manual reports.

Diagnoses full-stack failure modes in agent and LLM applications, including wrapper regressions, memory contamination, tool discipline failures, hidden repair loops, and rendering corruption. Produces severity-ranked findings and code-first fixes for production agent systems.

Use this skill to deeply inspect and audit the entire process of local AI coding-agent sessions. It tracks critical operational metrics such as token consumption, cost spikes, tool failures, latency anomalies, and overall health scores. Ideal for debugging unreliable code runs, establishing CI/CD health gates, or comparing different attempts to ensure semantic accuracy.