ClawHub
Back to skill

Security audit

Dataify DuckDuckGo Search

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed, read-only DuckDuckGo search helper that routes queries through Dataify using a Dataify API token.

Install only if you are comfortable sending search queries and selected search parameters to Dataify and using a Dataify API token. Do not include secrets or private data in search requests, and prefer raw JSON or HTML output only when you explicitly need it.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding
The skill instructs the agent to execute a Python script, access an API token from the environment, and perform outbound network requests, but it declares no corresponding permissions. This under-disclosure weakens policy enforcement and user awareness, making it easier for a seemingly simple search skill to access credentials and external services without explicit consent boundaries.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The skill presents itself as a DuckDuckGo search helper, but its documented behavior includes use of a third-party scraping API, credential handling, token setup guidance, and promotional account flows. That mismatch is security-relevant because users may consent to a low-risk web search while the skill actually routes queries to another provider and interacts with secrets and account onboarding.

Intent-Code Divergence

Medium
Confidence
81% confidence
Finding
The skill says to return script stdout directly without filtering or processing, but elsewhere says to present compact, user-facing results and only return raw JSON/HTML on explicit request. Conflicting output rules are dangerous because they can cause accidental disclosure of raw responses, metadata, internal envelopes, or other unintended content depending on which instruction an agent follows.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill directs the agent to pass the user's full request to an external Python script and submit it as form data to a third-party service, but it does not clearly warn that the complete prompt content may leave the local environment. In an agent context, users may include sensitive data in natural-language requests, creating a real risk of unintended data exfiltration to the external search provider.

Static analysis

No suspicious patterns detected.