ClawHub
Back to skill

Security audit

Gog(免费)

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple Google Workspace command guide whose Gmail and Sheets access matches its stated purpose, though users should handle OAuth credentials and returned data carefully.

Install only if you are comfortable authorizing a Google account for Gmail search and Sheets read/append. Use a least-privilege account where possible, review Sheet append commands before running them, avoid exporting unrelated API keys, and do not send mailbox or spreadsheet results to callback URLs you do not control.

SkillSpector

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

Medium
Confidence
92% confidence
Finding
The skill claims Gmail/Sheets operations are authenticated via OAuth credentials, but it also instructs users to export a generic API_KEY. This mismatch can cause users to provision unnecessary secrets, confuse the trust boundary, and increase the chance that unrelated API credentials are stored, exposed, or used incorrectly in automation.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The documented core operations access Gmail content and read or append data in Google Sheets, but the skill does not warn that these actions may expose sensitive mailbox contents or modify remote records. In an agent context, missing disclosure can lead users to authorize or invoke privacy-impacting actions without understanding the data access and write implications.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The callback_url parameter enables outbound network communication but is described with no privacy, SSRF, or data disclosure warning. Users may provide arbitrary endpoints without understanding that task results or metadata could be sent to third-party servers, potentially leaking sensitive information.

Static analysis

No suspicious patterns detected.