Back to skill

Security audit

微信小游戏本地化Skill

Security checks for vulnerabilities and agentic risk

Overview

This localization skill is useful but needs review because it can upload project images, read stored MCP credentials, install packages, and rewrite project files with insufficient containment.

Review before installing. Use only on trusted projects, prefer an isolated workspace, keep backups, avoid enabling image translation unless you are comfortable uploading project assets, provide MCP credentials explicitly over HTTPS, and avoid copying tokens into every IDE config until the path validation and HTTP handling are hardened.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Findings (7)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
scripts/upload-images.js:202
Finding

Arbitrary Local File Read and Exfiltration Through Image Path Traversal

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/upload-images.js:220
Finding

Shell Command Injection Through the Project Path During ZIP Creation

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/replace-text.js:125
Finding

Arbitrary File Modification and Command Injection in Text Replacement

Content
View full analysis
0 && !DRY_RUN) { fs.writeFileSync(fullPath, newContent, 'utf8'); const valid = validateFile(fullPath, filePath); if (!valid) { fs.writeFileSync(fullPath, originalContent, 'utf8'); } } } ``` JavaScript validation invokes a shell command containing the same path: ```js if (ext === '.js') { execSync(`node --check "${fullPath}"`, { stdio: 'pipe' }); return true; } ``` ### Technical Analysis `filePath` originates from translation metadata and is joined to `PROJECT_ROOT` without canonical containment enforcement. A malicious or corrupted `text_translations.json` can therefore specify paths containing `..` and cause the script to read, back up, modify, or restore files outside the project. If an external JavaScript filename includes shell metacharacters or embedded quotes, the path also reaches `execSync()` during syntax validation. This converts the path traversal primitive into a potential command-injection primitive. The backup path is constructed from the same untrusted relative path, so the backup operation does not establish a safe boundary. ### Attack P ...[truncated 890 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
scripts/replace-image.js:202
Finding

Arbitrary File Overwrite Through Image Replacement Path Traversal

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
references/mcp-image-translation.md:32
Finding

MCP Credentials and Project Images May Be Transmitted Over Plaintext HTTP

Content
View full analysis
", "TOKEN": "" }, "disabled": false } } } ``` The uploader explicitly permits either HTTP or HTTPS and sends credentials in request headers: ```js const url = new URL(MCP_URL); const isHttps = url.protocol === 'https:'; const transport = isHttps ? https : http; const reqOptions = { hostname: url.hostname, port: url.port || (isHttps ? 443 : 80), path: url.pathname + (url.search || ''), method: 'POST', headers: { 'Content-Type': 'application/json', 'APPID': APPID, 'TOKEN': TOKEN, 'Content-Length': Buffer.byteLength(body) } }; const req = transport.request(reqOptions, (res) => { // Response handling }); ``` ### Technical Analysis The documented endpoint uses unencrypted HTTP, and the implementation accepts that protocol without warning or rejection. The access token, AppID, image archives, tool calls, and server responses are therefore exposed to network observation and modification. HTTP provides neither confidentiality nor server authentication. An on-path attacker can read the token, replay it, replace responses, or modify uploaded data. ### Attack Path 1. A user follows the documented MCP configuration. 2. `upload-images.js` loads the HTTP URL and credentials. 3. The uploader sends `APPID` and `TOKEN` headers over an unencrypted connection. 4. A network observer captures the credentials and uploaded project data. 5. The observer can replay the token or alter MCP ...[truncated 397 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
scripts/upload-images.js:60
Finding

Overbroad Automatic Access to Multiple IDE Credential Stores

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
scripts/package.json:6
Finding

Unpinned Dependency Installation Without a Lockfile

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (83)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill advertises localization features but also performs remote upload of local images, reads user-level IDE MCP configuration files, and conducts chunked network communication with external services. Those behaviors materially change the trust and privacy profile of the skill, because local project assets and local configuration secrets may be exposed without sufficiently prominent disclosure at the trigger/overview level.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill advertises localization features but also performs remote upload of local images, reads user-level IDE MCP configuration files, and conducts chunked network communication with external services. Those behaviors materially change the trust and privacy profile of the skill, because local project assets and local configuration secrets may be exposed without sufficiently prominent disclosure at the trigger/overview level.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger conditions include broad generic terms such as '翻译' and '多语言', which can match ordinary conversation and activate a skill that performs scanning, dependency installation, file writes, and possible network uploads. Because this skill is stateful and mutating, overbroad invocation criteria materially increase the chance of accidental execution and unintended project changes.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The catch-all fuzzy trigger ('game for overseas users' and similar phrasing) lacks clear boundaries and can invoke a powerful workflow from ambiguous intent. In this context that is especially risky because the skill can install packages, read configs, upload images, and modify source files, so accidental activation has meaningful security and integrity consequences.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 50)May include surrounding context.

md
- `references/scan-analysis.md` — 扫描分析

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 298)May include surrounding context.

md
- `references/scan-analysis.md` — 扫描分析

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

md
- `references/execute-translation.md` — 执行翻译

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 351)May include surrounding context.

md
- `references/execute-translation.md` — 执行翻译

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 360)May include surrounding context.

md
- `references/execute-translation.md` — 执行翻译

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

md
- `references/apply-resources.md` — 应用本地化资源(文本替换、图片替换、语言包生成)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 372)May include surrounding context.

md
- `references/apply-resources.md` — 应用本地化资源(文本替换、图片替换、语言包生成)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 54)May include surrounding context.

md
- `references/quality-verification.md` — 本地化质量验证

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 396)May include surrounding context.

md
- `references/quality-verification.md` — 本地化质量验证

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 58)May include surrounding context.

md
- `scripts/scan-chinese.js` — **中文字符串全量扫描脚本 v1.0**(AST 扫描 JS/TS/JSON/CSV/TSV/WXML/HTML/CSS/Cocos/Unity/C#/XML/TXT,双引擎解析 esprima + @babel/parser 容错模式,精确行列 range

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 153)May include surrounding context.

md
- `scripts/scan-chinese.js` — **中文字符串全量扫描脚本 v1.0**(AST 扫描 JS/TS/JSON/CSV/TSV/WXML/HTML/CSS/Cocos/Unity/C#/XML/TXT,双引擎解析 esprima + @babel/parser 容错模式,精确行列 range

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 158)May include surrounding context.

md
- `scripts/scan-chinese.js` — **中文字符串全量扫描脚本 v1.0**(AST 扫描 JS/TS/JSON/CSV/TSV/WXML/HTML/CSS/Cocos/Unity/C#/XML/TXT,双引擎解析 esprima + @babel/parser 容错模式,精确行列 range

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 307)May include surrounding context.

md
- `scripts/scan-chinese.js` — **中文字符串全量扫描脚本 v1.0**(AST 扫描 JS/TS/JSON/CSV/TSV/WXML/HTML/CSS/Cocos/Unity/C#/XML/TXT,双引擎解析 esprima + @babel/parser 容错模式,精确行列 range

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 60)May include surrounding context.

md
- `scripts/replace-text.js` — 文本替换脚本 v3.5(含备份、dry-run、逐条即时语法校验、逐文件校验、最终全局校验、自动回滚、模板字符串增强兼容、TS 语法检查修复、CSV/TSV 数据文件替换支持、**精确字符串字面量定位** — 优先替换引号内的字符串,排除注释中的同名文本、**

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 375)May include surrounding context.

md
- `scripts/replace-text.js` — 文本替换脚本 v3.5(含备份、dry-run、逐条即时语法校验、逐文件校验、最终全局校验、自动回滚、模板字符串增强兼容、TS 语法检查修复、CSV/TSV 数据文件替换支持、**精确字符串字面量定位** — 优先替换引号内的字符串,排除注释中的同名文本、**

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 62)May include surrounding context.

md
- `scripts/generate-langpack.js` — 语言包生成脚本 v3.0(支持 Cocos Creator i18n 插件 / Cocos Creator L10N / LayaAir / Egret / Unity / 原生微信小游戏。自动部署语言包到引擎目录、自动注入 i18n 初始化代码、生

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 379)May include surrounding context.

md
- `scripts/generate-langpack.js` — 语言包生成脚本 v3.0(支持 Cocos Creator i18n 插件 / Cocos Creator L10N / LayaAir / Egret / Unity / 原生微信小游戏。自动部署语言包到引擎目录、自动注入 i18n 初始化代码、生

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 63)May include surrounding context.

md
- `scripts/verify-build-strings.js` — 编译产物二次复核脚本(基于 AST 扫描编译后 JS/JSON 中的残留中文,可选步骤)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 448)May include surrounding context.

md
- `scripts/verify-build-strings.js` — 编译产物二次复核脚本(基于 AST 扫描编译后 JS/JSON 中的残留中文,可选步骤)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 458)May include surrounding context.

md
- `scripts/verify-build-strings.js` — 编译产物二次复核脚本(基于 AST 扫描编译后 JS/JSON 中的残留中文,可选步骤)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 74)May include surrounding context.

md
- `references/scan-report-schema.md` — 扫描报告格式定义