Token导航 LogoToken导航

Agent Skill · 研究检索

solidity-security(坚固安全)

用于辅助安全审计、权限检查、凭据风险、认证流程和常见漏洞排查。它适合让 Agent 梳理敏感配置、检查依赖风险、分析鉴权逻辑或生成安全复核清单。使用时不能把工具输出直接当最终结论,涉及密钥、令牌、用户数据或生产系统时,应先确认最小权限、脱敏方式和操作边界。

solidity-security 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 37,724。

总安装 1,685周安装 75GitHub Stars 1下载 599核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:solidity-security(坚固安全)
来源仓库:https://github.com/0xlayerghost/solidity-agent-kit
仓库路径:skills/solidity-security
安装命令:
npx skills add https://github.com/0xlayerghost/solidity-agent-kit --skill solidity-security
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。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/0xlayerghost/solidity-agent-kit --skill solidity-security

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

适用于智能合约开发者的安全复核场景,solidity-security 属于研究检索类 Skill,可作为该场景下的辅助能力补充。可自动检测重入攻击、权限配置错误等关键风险点。通过分析代码结构和依赖关系,提供针对性的安全建议和改进方案。使用时需结合人工复核,确保关键操作符合最小权限原则。

安装后需将 .env 文件加入 .gitignore 防止密钥泄露。

来源仓库内路径:skills/solidity-security

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/0xlayerghost/solidity-agent-kit --skill solidi…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

solidity-security 怎么安装?

在终端执行 npx skills add https://github.com/0xlayerghost/solidity-agent-kit --skill solidity-securit…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

solidity-security 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

solidity-security 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

solidity-security 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

solidity-security 用的人多吗?

公开数据是累计安装 1,685 次,近一周安装 75 次,下载 599 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

solidity-security 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
solidity-security

坚固安全 · 本页

1,6851接触敏感数据未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Solidity Security Standards

Language Rule

  • Always respond in the same language the user is using. If the user asks in Chinese, respond in Chinese. If in English, respond in English.

Private Key Protection

  • Store private keys in .env, load via source.env — never pass keys as CLI arguments
  • Never expose private keys in logs, screenshots, conversations, or commits
  • Provide .env.example with placeholder values for team reference
  • Add .env to .gitignore — verify with git status before every commit

Security Decision Rules

When writing or reviewing Solidity code, apply these rules:

SituationRequired Action
External ETH/token transferUse ReentrancyGuard + Checks-Effects-Interactions (CEI) pattern
ERC20 token interactionUse SafeERC20 — call safeTransfer / safeTransferFrom, never raw transfer / transferFrom
Owner-only functionInherit Ownable2Step (preferred) or Ownable from OZ 4.9.x — Ownable2Step prevents accidental owner loss
Multi-role accessUse AccessControl from @openzeppelin/contracts/access/AccessControl.sol
Token approvalUse safeIncreaseAllowance / safeDecreaseAllowance from SafeERC20 — never raw approve
Price data neededUse Chainlink AggregatorV3Interface if feed exists; otherwise TWAP with min-liquidity check — never use spot pool price directly
Upgradeable contractPrefer UUPS (UUPSUpgradeable) over TransparentProxy; always use Initializable
Solidity version < 0.8.0Must use SafeMath — but strongly prefer upgrading to 0.8.20+
Emergency scenarioInherit Pausable, add whenNotPaused to user-facing functions; keep admin/emergency functions unpaused
Whitelist / airdropUse MerkleProof for gas-efficient verification — never store full address lists on-chain
Signature-based authUse ECDSA + EIP712 — never roll custom signature verification
Signature contentSignature must bind chainId + nonce + msg.sender + deadline — prevent replay and cross-chain reuse
Cross-chain bridge / third-party dependencyAudit all inherited third-party contract code — never assume dependencies are safe
Deprecated / legacy contractsPermanently pause or selfdestruct deprecated contracts — never leave unused contracts callable on-chain
UUPS upgrade pattern_authorizeUpgrade must have onlyOwner; implementation constructor calls _disableInitializers(); retain onlyProxy on upgradeTo — EVMbench/basin H-01
Multi-contract trust boundaryRouter/Registry relay calls must verify source contract authorization; never trust caller identity inside flash loan callbacks — EVMbench/noya H-08
Counter/ID + external callAll counter increments and ID assignments must complete before external calls; ETH refunds must be last — EVMbench/phi H-06

Reentrancy Protection

  • All contracts with external calls: inherit ReentrancyGuard, add nonReentrant modifier

- Import: @openzeppelin/contracts/security/ReentrancyGuard.sol (OZ 4.9.x)

  • Always apply CEI pattern even with ReentrancyGuard:

1. Checks — validate all conditions (require) 2. Effects — update state variables 3. Interactions — external calls last

Input Validation

  • Reject address(0) for all address parameters
  • Reject zero amounts for fund transfers
  • Validate array lengths match when processing paired arrays
  • Bound numeric inputs to reasonable ranges (prevent dust attacks, gas griefing)

Gas Control

  • Deployment commands must include --gas-limit (recommended >= 3,000,000)
  • Monitor gas with forge test --gas-report — review before every PR
  • Configure optimizer in foundry.toml: optimizer = true, optimizer_runs = 200
  • Avoid unbounded loops over dynamic arrays — use pagination or pull patterns

Pre-Audit Checklist

Before submitting code for review or audit, verify:

Access & Control:

  • All external/public functions have nonReentrant where applicable
  • No tx.origin used for authentication (use msg.sender)
  • No delegatecall to untrusted addresses
  • Owner transfer uses Ownable2Step (not Ownable) to prevent accidental loss
  • Contracts with user-facing functions inherit Pausable with pause() / unpause()
  • UUPS _authorizeUpgrade has onlyOwner modifier — EVMbench/basin H-01
  • Implementation constructor calls _disableInitializers() — EVMbench/basin H-01
  • Router/Registry relay operations verify source contract authorization — EVMbench/noya H-08

Token & Fund Safety:

  • All ERC20 interactions use SafeERC20 (safeTransfer / safeTransferFrom)
  • No raw token.transfer() or require(token.transfer()) patterns
  • Token approvals use safeIncreaseAllowance, not raw approve
  • All external call return values checked

Code Quality:

  • Events emitted for every state change
  • No hardcoded addresses — use config or constructor params
  • .env is in .gitignore

Oracle & Price (if applicable):

  • Price data sourced from Chainlink feed or TWAP — never raw spot price
  • Oracle has minimum liquidity check — revert if pool reserves too low
  • Price deviation circuit breaker in place

Testing:

  • forge test passes with zero failures
  • forge coverage shows adequate coverage on security-critical paths
  • Fuzz tests cover arithmetic edge cases (zero, max uint, boundary values)

Security Verification Commands

# Run all tests with gas report
forge test --gas-report

# Fuzz testing with higher runs for critical functions
forge test --fuzz-runs 10000

# Check test coverage
forge coverage

# Dry-run deployment to verify no runtime errors
forge script script/Deploy.s.sol --fork-url $RPC_URL -vvvv

# Static analysis (if slither installed locally)
slither src/

Slither MCP Integration (if available)

When slither MCP is configured, prefer it over the CLI for structured analysis:

ApproachWhen to Use
slither src/ (CLI)Quick local scan, raw terminal output
slither MCP get_detector_resultsStructured results with impact/confidence filtering, AI can parse and reason about findings
slither MCP get_contract_metadataUnderstanding contract structure before reviewing code
slither MCP get_function_sourceLocating exact function implementations faster than grep

Recommended: Use slither MCP when working with AI agents (results are structured and actionable). Use CLI for quick local checks during development.

Graceful degradation: If neither slither MCP nor slither CLI is available, rely on the Pre-Audit Checklist above and forge test --fuzz-runs 10000 for coverage.

平台分布

Codex

36.46%
按下载量换算218

Claude

32.84%
按下载量换算197

Cursor

16.75%
按下载量换算100

Gemini CLI

10.36%
按下载量换算62

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills