ClawHub
Back to skill

Security audit

小红书内容研究

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a coherent, read-only research workflow, with ordinary cautions about third-party searches and downloaded software.

Search terms are sent to SocialDataX, so use only information you may share with that service and keep your API key in supported secret storage. Prefer a reviewed, pinned CLI version with limited access to local files and credentials. This review covers the bundled instructions, not the downloaded package or future releases.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

MCP server referenced without pinned version: 'npx -y socialdatax-skills'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The command uses socialdatax-skills@latest, a mutable release tag, and npx -y allows execution without an installation confirmation. A compromised or malicious future package release could execute with the agent's permissions and access its API key or other accessible data; describing the skill as read-only does not constrain executable package code. This is a direct CLI invocation rather than an MCP server launch, but the unpinned dependency risk remains valid.

Content

No source excerpt is available for this finding.

The search workflow sends user-derived keywords to SocialDataX without a privacy disclosure.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The skill instructs the agent to execute a SocialDataX search using keywords derived from the user's actual research intent (line 51). Although it documents API-key handling and describes the skill as read-only, it does not warn that research queries, potentially including confidential topics or brand plans, are transmitted to a third-party service.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y socialdatax-skills'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The paginated search example repeats the mutable socialdatax-skills@latest dependency and automatic npx execution. If a package update is compromised, running this otherwise legitimate search can execute attacker-controlled code with access to the agent's environment and permissions. This is another occurrence of the same supply-chain weakness, not an independent attack mechanism.

Content

No source excerpt is available for this finding.