ClawHub
Back to skill

Security audit

Promptfoo

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent Promptfoo helper for local LLM evaluation and red-team testing, with expected but user-controlled use of external CLI commands and provider/API calls.

Before installing, review generated `promptfooconfig.yaml` files, avoid committing API keys or secrets, and run Promptfoo on small filtered suites first. Be aware that configured providers or HTTP targets may receive prompts, docs, test inputs, and outputs, and provider usage may cost money.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill instructs the agent to run Promptfoo CLI workflows that can write files (for example `init`, `generate`, red-team setup/report flows) and access the network, but the skill declares no permissions or constraints for those capabilities. This creates a mismatch between advertised safety boundaries and actual behavior, which can lead to unexpected file modification, outbound requests, or execution of external tooling without explicit authorization.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.