ClawHub
Back to skill

Security audit

小分子研发分析（Small-Molecule R&D Analysis）

Security checks across malware telemetry and agentic risk

Overview

This skill is purpose-aligned for molecule research, but it embeds a live third-party MCP endpoint with an API key and does not clearly disclose how user research queries are sent off-platform.

Review before installing. Treat molecule structures, CAS numbers, SMILES strings, target questions, and project context as potentially confidential because they may be sent to the referenced third-party MCP service. The embedded API key should be removed and rotated, and the skill should document external data handling before use in sensitive R&D workflows.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Context-Inappropriate Capability

High
Confidence
99% confidence
Finding
The skill embeds a live external MCP endpoint with a hard-coded API key directly in documentation. This exposes credentials to anyone who can read the skill and enables unauthorized use of the external service, while also creating an unreviewed data egress path for user queries to a third-party endpoint.

Context-Inappropriate Capability

High
Confidence
99% confidence
Finding
The repeated API-bearing endpoint compounds the credential exposure and increases the chance of accidental leakage, scraping, or downstream reuse. Because the skill instructs activation and use of that endpoint, it also operationalizes exfiltration of potentially sensitive user-provided research inputs to an external system without adequate trust controls.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The skill directs use of an external MCP endpoint but does not warn users that their compound identifiers, SMILES, CAS numbers, and research questions may be transmitted off-platform. In an R&D context, such queries can reveal confidential research interests, structures, screening targets, or proprietary molecule programs, making undisclosed data sharing a meaningful privacy and confidentiality risk.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.