ClawHub
Back to skill

Security audit

聚合搜索专业版

Security checks for vulnerabilities and agentic risk

Overview

The skill does not show malicious code, but it asks for command and file access while describing scheduled/background search and alert workflows that are not tightly scoped.

Review before installing. Use it only in a dedicated workspace, approve any exec, pip install, cron, API-server, or email-alert action explicitly, and keep search inputs and outputs away from secrets or unrelated local files. The artifact is not malicious, but its authority and background-task features are broader than a simple search helper.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
The skill is presented primarily as an internet search utility, but its documented core capabilities explicitly include broad file handling, API integration, and local command execution. That mismatch expands the trust boundary: a user invoking a search skill could unintentionally authorize arbitrary local actions unrelated to search, increasing the chance of misuse, unsafe command construction, or lateral abuse through the agent runtime.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The activation language is overly broad, covering generic data analysis, reporting, visualization, and automation scenarios well beyond a narrowly scoped search skill. In an agent ecosystem, vague triggers can cause over-invocation of a skill that has exec and file/output capabilities, leading to unnecessary exposure of local resources or unintended task execution.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The capability coverage section lists fragmented, vague, and keyword-like triggers that can match many unrelated user requests. Because the skill also advertises exporting, scheduling, and command execution, this broad matching increases the likelihood the agent invokes a higher-privilege skill in contexts where a narrower, safer tool should be used.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill documents file output, scheduled execution, archival, logging, and alerting behavior, but does not pair those side effects with prominent consent, scope, and safety controls. Unattended execution and file-writing features are risky in agent contexts because they can persist actions over time, overwrite local data, or transmit outputs without fresh user review.

Static analysis

No suspicious patterns detected.