ClawHub
Back to skill

Security audit

BugBug

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches a BugBug integration, but it under-discloses that it can execute BugBug tests while describing itself as for searching and reading data.

Review before installing if you expect a read-only BugBug helper. Only use it in workspaces where running BugBug tests is acceptable, and require explicit confirmation before invoking run_test or installing/signing into the oo CLI. VirusTotal and static scans were clean, and there is no evidence of deception or exfiltration.

SkillSpector

Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
92% confidence
Finding
The manifest says the skill is for 'searching and reading data,' but the documented actions include `run_test`, which initiates execution and changes external system state. This mismatch can mislead an agent or user into invoking a write-like operation under a read-only trust assumption, increasing the chance of unintended test execution.

Vague Triggers

Medium
Confidence
79% confidence
Finding
The instruction to use this skill for 'ANY BugBug request' is overly broad and can cause automatic invocation without sufficient task scoping or safety checks. In context, this is more dangerous because the skill exposes both read operations and a test-execution action, so broad routing may trigger higher-impact behavior than the description suggests.

Static analysis

No suspicious patterns detected.