xuziqiang98

基于LSP的智能代码安全审计Skill。利用OpenCode内置LSP进行代码理解、符号追踪、数据流分析，Agent自主识别项目特点并执行多维度安全审计（污点分析、认证授权、业务逻辑、敏感信息等），输出可验证的漏洞报告与PoC。适用于任何编程语言和项目类型。

Comprehensive binary exploitation techniques covering stack overflow, format string, heap exploitation, integer overflow, and advanced exploitation methods. Use when working on CTF challenges, binary vulnerability analysis, exploit development, or debugging memory corruption vulnerabilities in Linux binaries (x86/x64).

Comprehensive OpenCode SDK integration toolkit for client creation, session management, file operations, authentication, and real-time events. Use when Claude needs to build integrations with OpenCode server, generate SDK code, create automation scripts, or implement any functionality using the @opencode-ai/sdk package. Includes code generators, API reference, templates, and examples for all SDK workflows.

通用、语言/框架无关的离线静态污点分析审计 Skill。用于在只读前提下对任意仓库执行入口/Source 发现、Sink 扫描、污点策略建模、反向与正向数据流追踪、净化器与路径敏感校验、业务逻辑与越权审计、交叉验证、攻击链组合和最终 Markdown 报告输出，并落盘可复核中间产物。

专门用于在对话结束时，根据用户反馈和对话内容总结优化并迭代现有 Skills 的核心工具。它通过吸取对话中的“精华”（如成功的解决方案、失败的教训、特定的代码规范）来持续演进 Skills 库。