Back to skill

Security audit

Threads数据助手

Security checks for vulnerabilities and agentic risk

Overview

ClawScan’s verdict is benign with medium confidence: the declared purpose is public Threads querying, but scanners flag compliance records, upgrade guidance, an unpinned installer, and undisclosed sharing of URLs with a third party; coverage is partial.

The verdict is benign, but consider the reported compliance-record behavior, unpinned installer, error-triggered upgrades, and third-party URL sharing before installing. Skillspector marked its analysis partial: it inspected all 16 listed components, but some local references could not be resolved. The findings therefore do not provide complete coverage.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (12)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Local bookkeeping alone could be a supporting implementation detail. However, this code unconditionally marks legal terms as confirmed and logs a first-time confirmation, without any interaction or validation. That is a substantive, undeclared compliance-state change rather than ordinary support for the described read-only social-content queries.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Ordinary supporting logs would not constitute a mismatch. This code, however, specifically records a completed legal review, an undeclared compliance action unrelated to the described social-content queries. The mismatch is the unsupported compliance assertion, not merely the use of local filesystem writes or the absence of API calls in this limited chunk.

Content

No source excerpt is available for this finding.

Consent validation requires Chinese wording and explicitly rejects English “yes,” without offering a language choice.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The accepted confirmations are exclusively Chinese, while an unambiguous English “yes” is classified as vague and rejected. Although the framework cites Chinese laws, it does not explain why consent must be expressed in Chinese or offer users an alternative language.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skillhub'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The installation command can download and execute the currently resolved skillhub npm package without fixing the version. A compromised package release could execute with the installing user's permissions; although the finding incorrectly calls this an MCP server, the underlying supply-chain exposure is real.

Content

No source excerpt is available for this finding.

The compliance gate requires Chinese-language confirmation without offering a language choice.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The accepted confirmations are exclusively Chinese, while English responses such as yes are explicitly rejected and users are directed to enter Chinese wording. This imposes a language requirement without user opt-in or a documented justification for restricting consent to Chinese.

Content

No source excerpt is available for this finding.

The read-only query skill recommends modifying its installed code in response to API errors.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The manifest advertises read-only Threads querying and analysis (L005, L011), but L457–L459 prescribe skillhub upgrade threads-pro, including a forced upgrade, when endpoints return persistent errors. Updating the installed skill changes executable instructions rather than querying social data, and the error-triggered recommendations do not require a separate user request or explicit approval for that modification.

Content

No source excerpt is available for this finding.

The URL-to-post workflow forwards user-provided URLs to a third-party service without a privacy disclosure.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

L029 instructs the agent to pass the supplied URL directly to fetch_post_detail_v2, whose service base URL is https://www.aconfig.cn (L003), rather than Threads itself. The file does not warn users that their supplied URLs and lookup parameters are transmitted to this third-party provider; this is an executable workflow recipe, not merely an endpoint listing.

Content

No source excerpt is available for this finding.

The trigger “找人” (“find someone”) is overly broad and does not require a Threads-specific context.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

“找人” is a common everyday request that can refer to locating a contact, colleague, or person outside Threads. Although “Threads用户” is another listed trigger, the recipe does not require that platform context for the broader alternatives or provide exclusions.

Content

No source excerpt is available for this finding.

The confirmation rules require Chinese-language consent without offering a language choice.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

Every accepted confirmation is Chinese, while the English affirmation 'yes' is explicitly rejected. Together with the instruction to enter 同意, this imposes a language-specific interaction requirement without user opt-in or a documented reason why consent must be expressed in Chinese.

Content

No source excerpt is available for this finding.

The mandatory unsupported-capability response fixes the user's reply language to Chinese without offering a language choice.

Low
Category
Output Handling
Confidence
91% confidence
Finding

The instruction tells the agent to reply directly with a specific Chinese sentence when a request is unsupported. Unlike merely documenting the skill in Chinese, this prescribes user-facing output in Chinese without user opt-in or a documented locale-specific justification.

Content

No source excerpt is available for this finding.

The trigger “找人” (“find someone”) can activate a profile-search recipe without a platform-specific request.

Low
Category
Trigger Abuse
Confidence
94% confidence
Finding

“找人” is a common everyday phrase that does not necessarily express an intent to search Threads profiles. The matching rules resolve competing recipe matches but do not require Threads context or exclude unrelated people-finding requests.

Content

No source excerpt is available for this finding.

Batch-operation warnings can be triggered by ordinary words unrelated to bulk data collection.

Low
Category
Trigger Abuse
Confidence
96% confidence
Finding

The keyword list includes 全部 ('all'), 所有 ('every/all'), and 循环 ('loop') without requiring a data-collection action or providing exclusions. These words can occur in ordinary Threads-related requests that do not involve bulk collection, causing unintended compliance prompts.

Content

No source excerpt is available for this finding.