Back to skill

Security audit

X数据助手

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign with medium confidence: the scanners report a skill described as public Twitter/X analysis, with some related privacy and consent concerns, but the review was partial and found an unpinned installer and local compliance-record writes.

Before installing, consider whether you are comfortable with search terms and account identifiers being sent to the reported third-party service, and with the installer being unpinned. The review also reports local compliance-record writes and a balance-query capability outside the declared purpose. The verdict is benign, but Skillspector marked its analysis partial and noted missing or ambiguous references, so the evidence does not cover every referenced path or installed content.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (11)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The mismatch is not merely incidental logging or the absence of API calls in a partial snippet. The code unconditionally asserts that legal confirmation occurred, overwriting the confirmation marker and appending a corresponding audit event on each execution. Creating this consent-related state is materially different from the declared read-only data-analysis purpose and is not disclosed in the description.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The only supplied executable behavior is npx skillhub install x-pro, which invokes an installer rather than implementing the described API workflows. Installation is not disclosed in the description. The installed skill might implement the declared functionality, but its contents are not supplied and cannot be verified from this chunk.

Content

No source excerpt is available for this finding.

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

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The accepted confirmations are exclusively Chinese, while English "yes" is explicitly rejected and execution remains blocked until an accepted confirmation is supplied. Although the framework references Chinese laws, it does not explain why consent must be expressed in Chinese rather than an equally explicit affirmation in the user's language.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The installation command invokes an unversioned npm CLI through npx, allowing the retrieved executable to change between installations. A compromised package release or publisher account could execute attacker-controlled code with the installing user's privileges; this is a supply-chain exposure, although the referenced tool is an installer rather than an MCP server.

Content

No source excerpt is available for this finding.

The mandatory consent gate requires Chinese-language confirmation without user opt-in.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The gate accepts only listed Chinese affirmations and explicitly rejects English “yes,” instructing users to use wording such as “同意.” This imposes a Chinese-language requirement without offering a language choice or documenting why consent must be expressed in Chinese.

Content

No source excerpt is available for this finding.

Search workflows transmit user-provided keywords to a third-party service without a privacy warning.

Medium
Category
Output Handling
Confidence
89% confidence
Finding

The workflow directs the agent to use user-provided search keywords with an endpoint hosted at www.aconfig.cn (L003), rather than directly with Twitter/X. Although the API destination is documented, the file does not warn users that their search terms will be disclosed to this third-party provider.

Content

No source excerpt is available for this finding.

Lookup workflows lack a user-facing warning that Twitter/X identifiers are sent to a third-party service.

Medium
Category
Output Handling
Confidence
85% confidence
Finding

These chain recipes direct API lookups using account identifiers, while L003 sets the destination to www.aconfig.cn rather than Twitter/X. Although the service URL is documented, the file does not instruct the agent to disclose that requested usernames or account IDs will be shared with that third party, potentially revealing the user's lookup interests.

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
88% confidence
Finding

Every accepted confirmation is Chinese, while English 'yes' is explicitly rejected and the fallback prompt asks for Chinese wording. This imposes a language requirement on the user; the references to Chinese compliance laws do not explain why consent must be expressed only in Chinese.

Content

No source excerpt is available for this finding.

L107–L113 authorize a MaxHub account-balance query unrelated to the declared public Twitter/X analytics scope.

Low
Category
Excessive Agency
Confidence
93% confidence
Finding

The whitelist permits GET /api/maxhub/balance, which queries service-account balance information rather than public Twitter/X content or interactions. Account-balance access is not declared in the manifest and is not necessary to perform the described social-data queries.

Content

No source excerpt is available for this finding.

The social-circle recipe uses broad everyday phrases as activation triggers.

Low
Category
Trigger Abuse
Confidence
88% confidence
Finding

The semantic triggers “社交关系” (social relationships) and “社交圈” (social circle) can occur in ordinary discussions about personal relationships, without requesting account lookup. The activation condition does not require an explicit request to retrieve an account’s followers or followings, so unrelated conversation could activate this recipe.

Content

No source excerpt is available for this finding.

Batch-operation triggers include everyday words without requiring a data-collection context.

Low
Category
Trigger Abuse
Confidence
91% confidence
Finding

The keyword list includes broad terms such as 全部 ('all'), 所有 ('every'), and 循环 ('loop'). Although the description refers to bulk data collection, the matching rules do not explicitly require that context, so unrelated requests can trigger the batch-collection consent flow.

Content

No source excerpt is available for this finding.