Token导航 LogoToken导航TokenDH.com
开发执行命令github未标认证来源可访问clear审计通过

memory-safety-patterns内存安全模式

Agent Skill

memory-safety-patterns 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

157,248

周安装

6,254

GitHub Stars

34,529

下载量

50,904
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:memory-safety-patterns(内存安全模式)
来源仓库:https://github.com/wshobson/agents
仓库路径:skills/memory-safety-patterns
安装命令:
npx skills add https://github.com/wshobson/agents --skill memory-safety-patterns
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/wshobson/agents --skill memory-safety-patterns

简介

适用于 Rust、C++ 和 C 的 RAII、所有权、智能指针和资源管理的内存安全编程模式。

  • 涵盖六种核心内存错误类别(释放后使用、双重释放、泄漏、缓冲区溢出、悬空指针、数据竞争)以及特定于语言的预防策略
  • 提供 C++ 中的 RAII 模式以及析构函数、锁保护和事务;带有自定义删除器的智能指针指导(unique_ptr、shared_ptr、weak_ptr)
  • 实现 Rust 所有权、借用、生命周期和内部可变性(Cell、RefCell、Rc、Arc),并提供编译时安全保证
  • 包括使用 goto 清理的安全 C 模式、具有创建/销毁对的不透明指针以及 GCC 清理属性
  • 演示通过 C++ 中的容器和跨度以及 Rust 中的迭代器进行边界检查;跨语言使用原子、互斥体和 RwLock 的线程安全模式

SKILL.md

Memory Safety Patterns

Cross-language patterns for memory-safe programming including RAII, ownership, smart pointers, and resource management.

When to Use This Skill

  • Writing memory-safe systems code
  • Managing resources (files, sockets, memory)
  • Preventing use-after-free and leaks
  • Implementing RAII patterns
  • Choosing between languages for safety
  • Debugging memory issues

Core Concepts

1. Memory Bug Categories

Bug TypeDescriptionPrevention
Use-after-freeAccess freed memoryOwnership, RAII
Double-freeFree same memory twiceSmart pointers
Memory leakNever free memoryRAII, GC
Buffer overflowWrite past buffer endBounds checking
Dangling pointerPointer to freed memoryLifetime tracking
Data raceConcurrent unsynchronized accessOwnership, Sync

2. Safety Spectrum

Manual (C) → Smart Pointers (C++) → Ownership (Rust) → GC (Go, Java)
Less safe                                              More safe
More control                                           Less control

Patterns by Language

Pattern 1: RAII in C++

// RAII: Resource Acquisition Is Initialization
// Resource lifetime tied to object lifetime

#include <memory>
#include <fstream>
#include <mutex>

// File handle with RAII
class FileHandle {
public:
    explicit FileHandle(const std::string& path)
        : file_(path) {
        if (!file_.is_open()) {
            throw std::runtime_error("Failed to open file");
        }
    }

    // Destructor automatically closes file
    ~FileHandle() = default; // fstream closes in its destructor

    // Delete copy (prevent double-close)
    FileHandle(const FileHandle&) = delete;
    FileHandle& operator=(const FileHandle&) = delete;

    // Allow move
    FileHandle(FileHandle&&) = default;
    FileHandle& operator=(FileHandle&&) = default;

    void write(const std::string& data) {
        file_ << data;
    }

private:
    std::fstream file_;
};

// Lock guard (RAII for mutexes)
class Database {
public:
    void update(const std::string& key, const std::string& value) {
        std::lock_guard<std::mutex> lock(mutex_); // Released on scope exit
        data_[key] = value;
    }

    std::string get(const std::string& key) {
        std::shared_lock<std::shared_mutex> lock(shared_mutex_);
        return data_[key];
    }

private:
    std::mutex mutex_;
    std::shared_mutex shared_mutex_;
    std::map<std::string, std::string> data_;
};

// Transaction with rollback (RAII)
template<typename T>
class Transaction {
public:
    explicit Transaction(T& target)
        : target_(target), backup_(target), committed_(false) {}

    ~Transaction() {
        if (!committed_) {
            target_ = backup_; // Rollback
        }
    }

    void commit() { committed_ = true; }

    T& get() { return target_; }

private:
    T& target_;
    T backup_;
    bool committed_;
};

Pattern 2: Smart Pointers in C++

#include <memory>

// unique_ptr: Single ownership
class Engine {
public:
    void start() { /* ... */ }
};

class Car {
public:
    Car() : engine_(std::make_unique<Engine>()) {}

    void start() {
        engine_->start();
    }

    // Transfer ownership
    std::unique_ptr<Engine> extractEngine() {
        return std::move(engine_);
    }

private:
    std::unique_ptr<Engine> engine_;
};

// shared_ptr: Shared ownership
class Node {
public:
    std::string data;
    std::shared_ptr<Node> next;

    // Use weak_ptr to break cycles
    std::weak_ptr<Node> parent;
};

void sharedPtrExample() {
    auto node1 = std::make_shared<Node>();
    auto node2 = std::make_shared<Node>();

    node1->next = node2;
    node2->parent = node1; // Weak reference prevents cycle

    // Access weak_ptr
    if (auto parent = node2->parent.lock()) {
        // parent is valid shared_ptr
    }
}

// Custom deleter for resources
class Socket {
public:
    static void close(int* fd) {
        if (fd && *fd >= 0) {
            ::close(*fd);
            delete fd;
        }
    }
};

auto createSocket() {
    int fd = socket(AF_INET, SOCK_STREAM, 0);
    return std::unique_ptr<int, decltype(&Socket::close)>(
        new int(fd),
        &Socket::close
    );
}

// make_unique/make_shared best practices
void bestPractices() {
    // Good: Exception safe, single allocation
    auto ptr = std::make_shared<Widget>();

    // Bad: Two allocations, not exception safe
    std::shared_ptr<Widget> ptr2(new Widget());

    // For arrays
    auto arr = std::make_unique<int[]>(10);
}

Pattern 3: Ownership in Rust

// Move semantics (default)
fn move_example() {
    let s1 = String::from("hello");
    let s2 = s1; // s1 is MOVED, no longer valid

    // println!("{}", s1); // Compile error!
    println!("{}", s2);
}

// Borrowing (references)
fn borrow_example() {
    let s = String::from("hello");

    // Immutable borrow (multiple allowed)
    let len = calculate_length(&s);
    println!("{} has length {}", s, len);

    // Mutable borrow (only one allowed)
    let mut s = String::from("hello");
    change(&mut s);
}

fn calculate_length(s: &String) -> usize {
    s.len()
} // s goes out of scope, but doesn't drop since borrowed

fn change(s: &mut String) {
    s.push_str(", world");
}

// Lifetimes: Compiler tracks reference validity
fn longest<'a>(x: &'a str, y: &'a str) -> &'a str {
    if x.len() > y.len() { x } else { y }
}

// Struct with references needs lifetime annotation
struct ImportantExcerpt<'a> {
    part: &'a str,
}

impl<'a> ImportantExcerpt<'a> {
    fn level(&self) -> i32 {
        3
    }

    // Lifetime elision: compiler infers 'a for &self
    fn announce_and_return_part(&self, announcement: &str) -> &str {
        println!("Attention: {}", announcement);
        self.part
    }
}

// Interior mutability
use std::cell::{Cell, RefCell};
use std::rc::Rc;

struct Stats {
    count: Cell<i32>,           // Copy types
    data: RefCell<Vec<String>>, // Non-Copy types
}

impl Stats {
    fn increment(&self) {
        self.count.set(self.count.get() + 1);
    }

    fn add_data(&self, item: String) {
        self.data.borrow_mut().push(item);
    }
}

// Rc for shared ownership (single-threaded)
fn rc_example() {
    let data = Rc::new(vec![1, 2, 3]);
    let data2 = Rc::clone(&data); // Increment reference count

    println!("Count: {}", Rc::strong_count(&data)); // 2
}

// Arc for shared ownership (thread-safe)
use std::sync::Arc;
use std::thread;

fn arc_example() {
    let data = Arc::new(vec![1, 2, 3]);

    let handles: Vec<_> = (0..3)
        .map(|_| {
            let data = Arc::clone(&data);
            thread::spawn(move || {
                println!("{:?}", data);
            })
        })
        .collect();

    for handle in handles {
        handle.join().unwrap();
    }
}

Pattern 4: Safe Resource Management in C

// C doesn't have RAII, but we can use patterns

#include <stdlib.h>
#include <stdio.h>

// Pattern: goto cleanup
int process_file(const char* path) {
    FILE* file = NULL;
    char* buffer = NULL;
    int result = -1;

    file = fopen(path, "r");
    if (!file) {
        goto cleanup;
    }

    buffer = malloc(1024);
    if (!buffer) {
        goto cleanup;
    }

    // Process file...
    result = 0;

cleanup:
    if (buffer) free(buffer);
    if (file) fclose(file);
    return result;
}

// Pattern: Opaque pointer with create/destroy
typedef struct Context Context;

Context* context_create(void);
void context_destroy(Context* ctx);
int context_process(Context* ctx, const char* data);

// Implementation
struct Context {
    int* data;
    size_t size;
    FILE* log;
};

Context* context_create(void) {
    Context* ctx = calloc(1, sizeof(Context));
    if (!ctx) return NULL;

    ctx->data = malloc(100 * sizeof(int));
    if (!ctx->data) {
        free(ctx);
        return NULL;
    }

    ctx->log = fopen("log.txt", "w");
    if (!ctx->log) {
        free(ctx->data);
        free(ctx);
        return NULL;
    }

    return ctx;
}

void context_destroy(Context* ctx) {
    if (ctx) {
        if (ctx->log) fclose(ctx->log);
        if (ctx->data) free(ctx->data);
        free(ctx);
    }
}

// Pattern: Cleanup attribute (GCC/Clang extension)
#define AUTO_FREE __attribute__((cleanup(auto_free_func)))

void auto_free_func(void** ptr) {
    free(*ptr);
}

void auto_free_example(void) {
    AUTO_FREE char* buffer = malloc(1024);
    // buffer automatically freed at end of scope
}

Pattern 5: Bounds Checking

// C++: Use containers instead of raw arrays
#include <vector>
#include <array>
#include <span>

void safe_array_access() {
    std::vector<int> vec = {1, 2, 3, 4, 5};

    // Safe: throws std::out_of_range
    try {
        int val = vec.at(10);
    } catch (const std::out_of_range& e) {
        // Handle error
    }

    // Unsafe but faster (no bounds check)
    int val = vec[2];

    // Modern C++20: std::span for array views
    std::span<int> view(vec);
    // Iterators are bounds-safe
    for (int& x : view) {
        x *= 2;
    }
}

// Fixed-size arrays
void fixed_array() {
    std::array<int, 5> arr = {1, 2, 3, 4, 5};

    // Compile-time size known
    static_assert(arr.size() == 5);

    // Safe access
    int val = arr.at(2);
}
// Rust: Bounds checking by default

fn rust_bounds_checking() {
    let vec = vec![1, 2, 3, 4, 5];

    // Runtime bounds check (panics if out of bounds)
    let val = vec[2];

    // Explicit option (no panic)
    match vec.get(10) {
        Some(val) => println!("Got {}", val),
        None => println!("Index out of bounds"),
    }

    // Iterators (no bounds checking needed)
    for val in &vec {
        println!("{}", val);
    }

    // Slices are bounds-checked
    let slice = &vec[1..3]; // [2, 3]
}

Pattern 6: Preventing Data Races

// C++: Thread-safe shared state
#include <mutex>
#include <shared_mutex>
#include <atomic>

class ThreadSafeCounter {
public:
    void increment() {
        // Atomic operations
        count_.fetch_add(1, std::memory_order_relaxed);
    }

    int get() const {
        return count_.load(std::memory_order_relaxed);
    }

private:
    std::atomic<int> count_{0};
};

class ThreadSafeMap {
public:
    void write(const std::string& key, int value) {
        std::unique_lock lock(mutex_);
        data_[key] = value;
    }

    std::optional<int> read(const std::string& key) {
        std::shared_lock lock(mutex_);
        auto it = data_.find(key);
        if (it != data_.end()) {
            return it->second;
        }
        return std::nullopt;
    }

private:
    mutable std::shared_mutex mutex_;
    std::map<std::string, int> data_;
};
// Rust: Data race prevention at compile time

use std::sync::{Arc, Mutex, RwLock};
use std::sync::atomic::{AtomicI32, Ordering};
use std::thread;

// Atomic for simple types
fn atomic_example() {
    let counter = Arc::new(AtomicI32::new(0));

    let handles: Vec<_> = (0..10)
        .map(|_| {
            let counter = Arc::clone(&counter);
            thread::spawn(move || {
                counter.fetch_add(1, Ordering::SeqCst);
            })
        })
        .collect();

    for handle in handles {
        handle.join().unwrap();
    }

    println!("Counter: {}", counter.load(Ordering::SeqCst));
}

// Mutex for complex types
fn mutex_example() {
    let data = Arc::new(Mutex::new(vec![]));

    let handles: Vec<_> = (0..10)
        .map(|i| {
            let data = Arc::clone(&data);
            thread::spawn(move || {
                let mut vec = data.lock().unwrap();
                vec.push(i);
            })
        })
        .collect();

    for handle in handles {
        handle.join().unwrap();
    }
}

// RwLock for read-heavy workloads
fn rwlock_example() {
    let data = Arc::new(RwLock::new(HashMap::new()));

    // Multiple readers OK
    let read_guard = data.read().unwrap();

    // Writer blocks readers
    let write_guard = data.write().unwrap();
}

Best Practices

Do's

  • Prefer RAII - Tie resource lifetime to scope
  • Use smart pointers - Avoid raw pointers in C++
  • Understand ownership - Know who owns what
  • Check bounds - Use safe access methods
  • Use tools - AddressSanitizer, Valgrind, Miri

Don'ts

  • Don't use raw pointers - Unless interfacing with C
  • Don't return local references - Dangling pointer
  • Don't ignore compiler warnings - They catch bugs
  • Don't use unsafe carelessly - In Rust, minimize it
  • Don't assume thread safety - Be explicit

Debugging Tools

# AddressSanitizer (Clang/GCC)
clang++ -fsanitize=address -g source.cpp

# Valgrind
valgrind --leak-check=full ./program

# Rust Miri (undefined behavior detector)
cargo +nightly miri run

# ThreadSanitizer
clang++ -fsanitize=thread -g source.cpp

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

26.01%
按下载量换算13,240

Cursor

23.26%
按下载量换算11,840

OpenCode

17.53%
按下载量换算8,923

Gemini CLI

11.08%
按下载量换算5,640

Antigravity

8.05%
按下载量换算4,098

Codex

2.94%
按下载量换算1,497

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/wshobson/agents --skill memory-safety-patterns;npx skills add wshobson/agents --skill "memory-safety-patterns" 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。

来源信息

继续浏览同类 Skills