Back to skill

Security audit

GUAIKEI快手内容研究助手

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill benign with medium confidence: scanners report Kuaishou research features and no decisive malicious behavior, while noting privacy and incomplete-analysis limitations.

The verdict is benign, but confidence is medium and analysis was partial: 20 of 32 components were fully scanned, with 12 partially inspected. Before installing, consider the reported third-party transmission of queries and links, and note the possible gap between the keyword-search description and the additional retrieval capabilities.

Vulnerability Patterns
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared primary capability is keyword search, but this code has no keyword argument or search operation. Its required input is a creator profile URL or numeric user ID, and its core workflow invokes createPostTask and getPostTask to retrieve that creator's data and posts. Although these results could support the listed research use cases, the retrieval capability is materially different from the declared search function. Token authentication, validation, and output handling are supporting implementation details rather than separate mismatches.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared capability is keyword-based Kuaishou search, but this code takes a video URL or ID and a comment limit, then calls createCommentTask and getCommentTask. It accepts no search keyword and performs no keyword search. Although comments may support the broader research use cases listed in the description, targeted video-comment retrieval is a distinct, undeclared capability. Token handling, validation, and output logging are supporting implementation details, not separate grounds for the mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The code submits a URL and limit to /api/kuaishou/comment/url and polls /api/kuaishou/comment/info. It contains no keyword input or search operation. Although comments could support content research, retrieving comments from a specified URL is a distinct data-access capability not disclosed by the keyword-search description. Token handling, polling, and response validation are supporting implementation details, not separate mismatches.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied code accepts a URL, not a keyword, and uses post-specific task endpoints rather than implementing keyword search. This is a materially different retrieval operation, even though both could support content research. The task polling and response validation are ordinary implementation details, not separate mismatches. The code does not establish that it publishes posts.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The supplied commands target an individual video through comment-cli.js; neither performs a keyword search. Although comments could support content research, that broad use case does not disclose the distinct comment-retrieval capability. The CLI implementation is not provided, so the assessment is limited to the behavior indicated by these invocations.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The declared capability is keyword-based search, but the command targets a particular video through comment-cli.js and supplies no keyword-search input. Although comments could support the broader research use cases, standalone per-video comment retrieval is a distinct, undeclared capability. This assessment is limited to the visible invocation; the script's internal behavior cannot be verified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The supplied command targets a specific user profile and supplies no keyword or search query. Although profile posts could support the broader research use cases, this is a distinct retrieval capability not disclosed by the keyword-search description. The snippet does not establish whether any data is actually retrieved or whether it is real-time.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The supplied commands target a specific Kuaishou profile or identifier through post-cli.js and --url. This is a distinct retrieval mode from the explicitly declared keyword search, even though it could support similar research goals. The CLI implementation is not supplied, so its exact retrieval behavior and any additional side effects cannot be verified.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

md
- "帮我搜快手里「具身智能」的高赞视频 → node scripts/kuaishou/search-cli.js --keyword \"具身智能\" --sort 2 --limit 20"

Incomplete referenced artifact analysis

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
- "最近一周最新的快手 AI 教程,要 20 条 → node scripts/kuaishou/search-cli.js --keyword \"AI 教程\" --sort 1 --time 7 --limit 20"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 86)May include surrounding context.

md
| 找选题、看高赞 | `search-cli.js` | 视频数组。点赞、播放、评论是数字 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 123)May include surrounding context.

md
3. 有关键词,且没有上面两类信号 → `search-cli.js`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 154)May include surrounding context.

md
ut 为空,这是正常现象。短参数大小写都可以。含空格的关键词用双引号包成一个参数。关键词以 `-` 开头时写成 `--keyword=-AI`,或 `node scripts/kuaishou/search-cli.js -- -AI`。完整字段见 `assets/*_cli_req.schema.json` 与 [r

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 75)May include surrounding context.

md
- "这条快手视频评论区在说什么 → node scripts/kuaishou/comment-cli.js --url \"https://www.kuaishou.com/short-video/3xabc\" --limit 40"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 87)May include surrounding context.

md
| 看评论区 | `comment-cli.js` | 评论数组 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 121)May include surrounding context.

md
1. 要评论、留言、舆情,且对象是一条视频 → `comment-cli.js`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 130)May include surrounding context.

md
- `kuaishou.com/short-video/{id}`、`v.kuaishou.com/*`、裸的 `3x...` 只能给 `comment-cli.js`。搜索结果里用视频 `id` 或 `url`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 146)May include surrounding context.

md
| 评论区 | `comment-cli.js`。没说条数用默认 10 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

md
- "这个快手博主最近 20 条作品 → node scripts/kuaishou/post-cli.js --url \"https://www.kuaishou.com/profile/3xabc\" --sort 0 --limit 20"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
- "这个快手博主多少粉丝 → node scripts/kuaishou/post-cli.js --url 123456 --limit 0"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 122)May include surrounding context.

md
2. 要作品、主页、账号、博主、粉丝 → `post-cli.js`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 128)May include surrounding context.

md
- `kuaishou.com/profile/{id}` 只能给 `post-cli.js`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 129)May include surrounding context.

md
- 纯数字 `user_id` 只能给 `post-cli.js`。搜索结果里用 `user.user_id`,或直接用 `user.url`

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The quick-start workflow omits disclosure that user-supplied keywords and links are sent to a third-party service.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

The README instructs users to run searches and account/video queries, but only identifies guaikei.com as the place to obtain a token; it does not disclose that the supplied keywords or links are transmitted there. The blanket privacy assurance at L74 can further obscure this distinction: retrieving public content does not mean the user's research inputs are non-sensitive.

Content

No source excerpt is available for this finding.