ClawHub
Back to skill

Security audit

llm-provider AI工具

Security checks for vulnerabilities and agentic risk

Overview

This skill is mostly coherent for managing OpenAI-style provider resources, but it asks for broad agent permissions and gives examples that can upload local data or create remote resources without sufficiently tight scoping.

Install only if you intend agents to manage OpenAI-style provider resources through ClawLink. Before use, confirm any upload, batch, assistant, vector-store, image, or fine-tuning action, and avoid pointing it at private documents unless you are comfortable sending that data to the connected provider account.

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
93% confidence
Finding
The skill advertises itself for very broad situations like general AI model use, intelligent dialogue, agent orchestration, and LLM applications, which can cause an agent to invoke it for many unrelated requests. In this context, that overbroad routing is risky because the skill exposes read/exec/write capabilities and operations that can send data to an external provider, increasing the chance of unnecessary tool use or unintended data handling.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The examples show uploading local files and creating remote resources, but they do not prominently warn that file contents and prompts will be transmitted to a third-party provider and may persist in remote systems. This is dangerous because users or calling agents may treat these as routine file operations and inadvertently exfiltrate sensitive documents, embeddings data, prompts, or generated resources to an external service.

Static analysis

No suspicious patterns detected.