Cydew

Security checks across malware telemetry and agentic risk

Overview

This is a coherent API onboarding guide for creating a Cydew marketplace listing, with disclosed but real privacy and token-handling considerations.

Before using this skill, decide what contact details, portfolio links, pricing, and availability you are comfortable making discoverable. Use business contact information where possible, review every POST or PUT body before sending it, and handle Clerk M2M tokens as secrets with least-privilege claims.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Intent-Code Divergence

Medium

Confidence: 97% confidence
Finding: The documentation states verification is manual, then instructs users to set `isVerified` to `false` themselves if no verification flow exists. Even though the example uses `false`, this normalizes client control over a trust-state field that should be server-managed, creating a dangerous precedent that clients may also be able to set or tamper with verification status. In a marketplace context, trust badges materially affect buyer decisions, so ambiguity around ownership of this field can enable impersonation or fraudulent credibility signals if the API accepts it.

Missing User Warnings

Medium

Confidence: 85% confidence
Finding: The skill instructs agents to publish listings containing personal data such as name, email, bio, avatar, timezone, and availability details, but it does not warn that this information may become publicly discoverable or privacy-sensitive. This can lead users or downstream agents to over-share personal or identifying information without informed consent, increasing risks of scraping, profiling, spam, and targeted social engineering. The marketplace/search context makes the omission more dangerous because discoverability is an explicit goal of the workflow.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The authentication guidance tells users to use Clerk M2M tokens with `agentId` and `requesterId` claims, but it does not warn that these tokens are sensitive credentials or that authorization claims must be issued and validated by the trusted identity provider/server. This omission can encourage unsafe handling, logging, embedding, or even attempted self-construction of privileged claims, which could lead to account takeover or unauthorized access if the surrounding implementation is weak. Because the document covers owner-only endpoints and identity-bound actions, mishandling these tokens has direct security consequences.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal