Back to skill

Security audit

Linkedin数据助手

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill benign with medium confidence: its stated purpose and reported capabilities are broadly aligned, though scanners flag an unpinned installer and an automatic legal-confirmation record, and some references were unresolved.

The verdict is benign, not a guarantee of no risk. Before installing, consider whether you are comfortable using an unpinned installer and sending query details to the named third-party API; also note the reported automatic confirmation record and broad recipe triggers. Skillspector inspected all listed components, but coverage was marked partial because some local references were missing or ambiguous, so the evidence is incomplete.

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 Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (11)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

Local compliance logging could be a supporting implementation detail, but this code goes beyond logging: it unconditionally asserts that legal confirmation occurred. That changes compliance state and creates an unsupported acknowledgment record, neither of which is disclosed by the public-data-query description. The mismatch is the automatic confirmation assertion, not merely the local file writes or the absence of API calls in 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
95% confidence
Finding

The accepted confirmations are exclusively Chinese, while English “yes” is explicitly rejected and users cannot proceed without an accepted confirmation. Although the framework cites Chinese laws, it does not explain why consent must be expressed in Chinese or offer equivalent confirmations 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 executes an unpinned npm CLI, so its executable contents can change without a corresponding change to this documentation. A compromised future release could execute with the installing user's permissions and access local credentials; this is a supply-chain exposure, although the finding incorrectly describes the CLI as an MCP server.

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
97% confidence
Finding

The confirmation rules accept only enumerated Chinese phrases and explicitly reject English responses such as 'yes'. This forces users to confirm in Chinese without an opt-in or a documented reason for the language restriction; requiring informed consent does not itself justify restricting its language.

Content

No source excerpt is available for this finding.

The trigger "快速了解" ("quickly understand") can activate a user-profile recipe for unrelated requests.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

“快速了解” is an everyday phrase that can refer to any subject, but the index maps it to the user_top_card recipe without requiring a person, profile, or LinkedIn context. The longest-match and tie-breaking rules do not prevent an unrelated request containing this phrase from selecting that recipe.

Content

No source excerpt is available for this finding.

The trigger “快速了解” (“quickly understand”) is too broad for a user-profile retrieval recipe.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

This phrase commonly appears in requests to understand topics unrelated to a person or their profile. The activation instruction does not require a LinkedIn or user-profile context, so unrelated requests could trigger this recipe.

Content

No source excerpt is available for this finding.

The chain recipes send people-search terms and profile identifiers to a third-party API without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The chain recipes at L047–L049 direct searches and profile lookups through the external API hosted at www.aconfig.cn (L003), transmitting names, usernames, or profile URLs. Although the API host is listed, the file does not warn users that their search inputs and target-profile identifiers are disclosed to a service other than LinkedIn.

Content

No source excerpt is available for this finding.

The mandatory unsupported-request reply fixes the user-facing response in Chinese without a language choice.

Low
Category
Output Handling
Confidence
90% confidence
Finding

The instruction tells the agent to reply directly with a fixed Chinese sentence when a request is unsupported. It provides no option to translate that response into the user's language and gives no locale-specific justification.

Content

No source excerpt is available for this finding.

The standalone trigger “hashtag” does not distinguish requests to retrieve posts from other hashtag-related requests.

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The recipe treats a semantic match on “hashtag” as an activation condition for searching posts. This could also match requests to explain hashtags or suggest one for a draft, and no exclusion conditions distinguish those requests from retrieval intent.

Content

No source excerpt is available for this finding.

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

Low
Category
Trigger Abuse
Confidence
88% confidence
Finding

The keyword list includes 全部 ('all'), 所有 ('every/all'), and 循环 ('loop'), which can occur in ordinary requests unrelated to bulk data collection. Although the category description mentions bulk collection, the matching rules specify no contextual requirement or exclusions for these broad terms.

Content

No source excerpt is available for this finding.

Confirmation rules require Chinese consent wording without offering a language choice.

Low
Category
Output Handling
Confidence
96% confidence
Finding

Every accepted confirmation is Chinese, while English 'yes' is explicitly rejected. This imposes a language requirement on user consent without opt-in or a documented justification for restricting confirmation to Chinese; citing regional laws elsewhere does not explain that restriction.

Content

No source excerpt is available for this finding.