ClawHub
Back to skill

Security audit

Fox Cellcog

Security checks for vulnerabilities and agentic risk

Overview

This is a real CellCog cloud-integration skill, but it deserves Review because it can upload chosen local files, spend API credits, run background cloud tasks, and write remote-generated files to user-specified local paths without tight scoping in the artifact.

Install only if you trust CellCog and are comfortable sending selected files to its cloud service. Use a dedicated API key if possible, approve credit-consuming tasks deliberately, avoid secrets or regulated data unless CellCog is approved for that data, and keep generated outputs in a controlled directory rather than arbitrary workflow or startup paths.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
85% confidence
Finding
The skill repeatedly markets itself as a general-purpose sub-agent for 'quality work' and 'most tasks,' which is broad enough to trigger on many ordinary user requests. In an agent ecosystem, this can cause over-selection of a cloud-connected skill and increase the chance that sensitive data is routed to an external service without deliberate user intent.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The documentation explains that files inside <SHOW_FILE> are read by CellCog, but it does not present this as a clear up-front warning that those files are transmitted to a third-party cloud service. Because the skill encourages attaching arbitrary local files, users or upstream agents may disclose sensitive documents, recordings, or code without informed consent.

Static analysis

No suspicious patterns detected.