ClawHub
Back to skill

Security audit

S2S Forecasting Expert (FuXi, FengWu, AIFS)

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed local code-generation helper for S2S weather forecasting models, with no evidence of hidden execution, data exfiltration, or persistence.

Install this if you want an agent to help generate substantial local ML forecasting code. Review generated training and data pipeline code before running it, especially resource-heavy distributed training or ERA5-related workflows, but the skill package itself does not show hidden network, persistence, or execution behavior.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The skill advertises automatic invocation on broad phrases like building S2S models, CRPS training, AI weather model architecture, and ERA5 pipelines, which are loosely bounded enough to trigger on general ML or weather-engineering requests. While this is not a direct code-execution flaw, overbroad activation increases the chance the skill is invoked outside its narrow intended context, potentially steering users into generating complex local code or training scaffolds they did not explicitly request.

Static analysis

No suspicious patterns detected.