ClawHub
Back to skill

Security audit

CSP Policy Generator

Security checks for vulnerabilities and agentic risk

Overview

This skill gives coherent CSP-generation guidance and its network fetching examples are disclosed and purpose-aligned, though users should only point it at sites they intend to analyze.

Install only if you are comfortable with the agent making web requests to sites you specify for CSP analysis. Use it on domains you own or are authorized to test, and avoid pointing it at localhost, private network addresses, admin panels, or sensitive internal services.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

External Script Fetching

High
Category
Supply Chain
Content
```bash
# Fetch the page and extract all resource URLs
curl -sL "https://$HOST" | python3 -c "
import sys, re
from urllib.parse import urlparse
html = sys.stdin.read()
Confidence
90% confidence
Finding
The skill instructs fetching arbitrary remote content from a user-controlled host and immediately piping it into a Python parser. Although the fetched HTML is treated as data rather than executed as code, this still causes the agent to make network requests to attacker-influenced destinations, which can enable SSRF-style access to internal services, unexpected data exfiltration through outbound requests, or analysis of untrusted content without safety boundaries. In this skill’s context, the behavior is core functionality, but it is still security-relevant because the target host is variable and the agent is encouraged to retrieve live remote content automatically.

Static analysis

No suspicious patterns detected.