Token导航 LogoToken导航

Agent Skill · 研究检索

security-review(安全审查)

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。

security-review 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 61,129。

总安装 356周安装 15GitHub Stars 6下载 125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:security-review(安全审查)
来源仓库:https://github.com/troykelly/codex-skills
仓库路径:skills/security-review
安装命令:
npx skills add https://github.com/troykelly/codex-skills --skill security-review
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/troykelly/codex-skills --skill security-review

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。

安全审查技能用于辅助安全审计、权限检查和常见漏洞排查。security-review 属于研究检索类 Skill,可作为该场景下的辅助能力补充。适用于梳理敏感配置、分析依赖风险或生成安全复核清单的场景。使用时不能将工具输出直接当作最终结论,需结合人工判断。涉及密钥、令牌或生产系统时,应先确认最小权限和脱敏方式。

该技能归类于研究检索类别,适合安全评估和合规检查场景。

来源仓库内路径:skills/security-review

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/troykelly/codex-skills --skill security-review;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

security-review 怎么安装?

在终端执行 npx skills add https://github.com/troykelly/codex-skills --skill security-review。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

security-review 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。

security-review 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。

security-review 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。

security-review 用的人多吗?

公开数据是累计安装 356 次,近一周安装 15 次,下载 125 次,GitHub 6 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。

security-review 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
security-review

安全审查 · 本页

3566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

Security Review

Overview

Dedicated security review for code handling authentication, authorization, user input, APIs, databases, or credentials.

Core principle: Security issues require specialized attention beyond general code review.

Trigger: This review is MANDATORY when changes touch security-sensitive paths.

Announce at start: "I'm performing a security review of this code."

When Required

This skill is MANDATORY when ANY of these files are modified:

PatternExamples
**/auth/**src/auth/login.ts, lib/auth/session.js
**/security/**src/security/encryption.ts
**/middleware/**src/middleware/authenticate.ts
**/api/**src/api/endpoints.ts
**/*password*utils/passwordHash.ts
**/*token*services/tokenService.ts
**/*secret*config/secrets.ts
**/*credential*lib/credentials.js
**/*session*middleware/session.ts
**/routes/**src/routes/protected.ts
**/*.sqlmigrations/001_users.sql

Check with:

git diff --name-only HEAD~1 | grep -E '(auth|security|middleware|api|password|token|secret|credential|session|routes|\.sql)'

OWASP Top 10 Checklist

Review against each category:

1. Injection (A03:2021)

CheckVerify
SQL InjectionAll queries use parameterized statements
Command InjectionNo user input in shell commands, or properly escaped
LDAP InjectionLDAP queries use proper escaping
XPath InjectionXPath queries use parameterized approach
Template InjectionTemplate engines configured safely
// VULNERABLE
db.query(`SELECT * FROM users WHERE id = '${userId}'`);

// SECURE
db.query('SELECT * FROM users WHERE id = ?', [userId]);

2. Broken Authentication (A07:2021)

CheckVerify
Password StoragePasswords hashed with bcrypt/argon2 (not MD5/SHA1)
Session ManagementSecure, HttpOnly, SameSite cookies
Token HandlingJWTs signed, validated, short-lived
Brute Force ProtectionRate limiting on auth endpoints
Credential ExposureNo credentials in logs, errors, or responses

3. Sensitive Data Exposure (A02:2021)

CheckVerify
Data in TransitHTTPS enforced, TLS 1.2+
Data at RestSensitive data encrypted
Secrets in CodeNo hardcoded API keys, passwords, tokens
Error MessagesNo sensitive info in error responses
LoggingNo sensitive data logged
# Check for hardcoded secrets
grep -rE '(password|secret|api_key|token)\s*[:=]\s*["\047][^"\047]+["\047]' src/

4. XML External Entities (A05:2021)

CheckVerify
XML ParsingExternal entities disabled
DTD ProcessingDTD processing disabled if not needed

5. Broken Access Control (A01:2021)

CheckVerify
Authorization ChecksEvery endpoint verifies permissions
Direct Object ReferencesObject access validated against user
Privilege EscalationCannot elevate own privileges
CORSProperly restricted origins
Method RestrictionOnly allowed HTTP methods accepted

6. Security Misconfiguration (A05:2021)

CheckVerify
Default CredentialsNo default passwords in use
Error HandlingStack traces not exposed to users
Security HeadersCSP, X-Frame-Options, etc. set
Debug ModeDisabled in production
Unnecessary FeaturesUnused endpoints/features removed

7. Cross-Site Scripting (A03:2021)

CheckVerify
Output EncodingUser input encoded before display
DOM XSSinnerHTML not used with user input
Template SafetyTemplate engine auto-escapes
CSPContent Security Policy configured
// VULNERABLE
element.innerHTML = userInput;

// SECURE
element.textContent = userInput;
// OR
element.innerHTML = DOMPurify.sanitize(userInput);

8. Insecure Deserialization (A08:2021)

CheckVerify
Object DeserializationUntrusted data not deserialized
JSON ParsingSafe JSON.parse usage
Type ValidationDeserialized objects validated

9. Using Components with Known Vulnerabilities (A06:2021)

CheckVerify
Dependency Auditpnpm audit / pip audit clean
Outdated PackagesNo critically outdated dependencies
CVE CheckNo known CVEs in dependencies
# Run dependency audit
pnpm audit --prod
# or
pip-audit

10. Insufficient Logging & Monitoring (A09:2021)

CheckVerify
Auth EventsLogin success/failure logged
Access ControlPermission denials logged
Input ValidationValidation failures logged
Sensitive ActionsAdmin actions logged
Log IntegrityLogs protected from tampering

Review Process

Step 1: Identify Security-Sensitive Changes

# List changed files matching security patterns
git diff --name-only HEAD~1 | grep -E '(auth|security|middleware|api|password|token|secret|session)'

Step 2: Review Each File

For each security-sensitive file:

  1. Read the entire file (not just diff)
  2. Check against all 10 OWASP categories
  3. Note any findings with severity

Step 3: Check Dependencies

# Audit dependencies
pnpm audit --prod

# Check for outdated
pnpm outdated

Step 4: Document Findings

Use severity levels:

SeverityDescriptionAction
CRITICALExploitable vulnerability, data breach riskMUST fix before merge
HIGHSignificant security weaknessMUST fix before merge
MEDIUMDefense-in-depth issueSHOULD fix before merge
LOWMinor improvementMAY fix in future issue

Step 5: Add Security Section to Review Artifact

Add to the main review artifact:

### Security Review

**Security-Sensitive:** YES
**Reviewed By:** [WORKER_ID or codex-subagent security-reviewer]
**OWASP Categories Checked:** 10/10

#### Security Findings

| # | OWASP Category | Severity | Finding | Status |
|---|----------------|----------|---------|--------|
| 1 | A03 Injection | CRITICAL | SQL injection in findUser() | FIXED |
| 2 | A02 Sensitive Data | HIGH | API key in config.ts | DEFERRED #456 |
| 3 | A01 Access Control | MEDIUM | Missing auth on /admin | FIXED |

#### Dependency Audit

pnpm audit: 0 vulnerabilities

**Security Review Status:** [PASS|ISSUES_FIXED|ISSUES_DEFERRED]

Security Review Artifact (Standalone)

If security review is extensive, post as separate comment:

<!-- SECURITY_REVIEW:START -->
## Security Review

| Property | Value |
|----------|-------|
| Issue | #123 |
| Reviewer | `security-reviewer` subagent (codex-subagent) |
| Reviewed | 2025-12-29T10:30:00Z |

### Files Reviewed

- src/auth/login.ts
- src/middleware/authenticate.ts
- src/api/users.ts

### OWASP Checklist Results

| # | Category | Status | Notes |
|---|----------|--------|-------|
| A01 | Broken Access Control | PASS | Auth middleware on all protected routes |
| A02 | Cryptographic Failures | PASS | bcrypt for passwords, TLS enforced |
| A03 | Injection | FIXED | Parameterized SQL queries now |
| A04 | Insecure Design | PASS | - |
| A05 | Security Misconfiguration | PASS | - |
| A06 | Vulnerable Components | PASS | pnpm audit clean |
| A07 | Auth Failures | PASS | Rate limiting, secure sessions |
| A08 | Data Integrity Failures | PASS | - |
| A09 | Logging Failures | NOTE | Consider adding auth failure logging |
| A10 | SSRF | N/A | No server-side requests |

### Dependency Audit

found 0 vulnerabilities

**Security Review Status:** PASS
<!-- SECURITY_REVIEW:END -->

Checklist

Before completing security review:

  • All changed security-sensitive files reviewed
  • All 10 OWASP categories checked
  • Dependency audit run
  • All CRITICAL/HIGH findings fixed
  • Any deferred findings have tracking issues
  • Security section added to review artifact
  • Security-Sensitive marked YES in main artifact

Integration

This skill is triggered by:

  • Changes to security-sensitive file patterns
  • Explicit invocation
  • codex-subagent security-reviewer

This skill integrates with:

  • comprehensive-review - Security is criterion #4
  • review-gate - Verifies security review for sensitive changes

This skill is enforced by:

  • .codex/rules/security-sensitive.md conditional rules
  • PR gate automation (block PR if security-sensitive without review)

平台分布

Codex

34.63%
按下载量换算43

Claude

31.92%
按下载量换算40

Cursor

17.69%
按下载量换算22

Gemini CLI

8.37%
按下载量换算10

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