ClawHub
Back to skill

Security audit

crawlora-datasets

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly a disclosed Crawlora dataset lookup tool, but its helper script exposes broader Crawlora API access than the skill description promises.

Review this before installing if you expect a dataset-only skill. It requires your Crawlora API key and the included helper can be used to call broader Crawlora API paths, not just the documented hosted dataset endpoints. There is no evidence of hidden persistence or local data theft, but the scope should be tightened or clearly disclosed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (2)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The helper accepts an arbitrary API path as its first positional argument and forwards it directly to the Crawlora API, which means the skill can access endpoints beyond the hosted-dataset query scope described in the metadata. In this context, the danger is capability expansion and policy bypass: an agent or user can invoke unrelated or more sensitive Crawlora functionality than the skill claims to expose.

Intent-Code Divergence

Medium
Confidence
90% confidence
Finding
The usage examples advertise endpoints such as /amazon/search, /youtube/transcript, /google/search, and /google/trends that do not align with the stated purpose of querying pre-built hosted datasets. This mismatch increases the likelihood that an agent will use the skill for live-content retrieval or broader API access than intended, undermining least-privilege expectations.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical
Code
suspicious.exposed_secret_literal
Location
SKILL.md:38