ClawHub
Back to skill

Security audit

Json Yaml Converter Free

Security checks across malware telemetry and agentic risk

Overview

This is a local JSON/YAML conversion instruction skill with some inconsistent and overbroad documentation, but no artifact-backed hidden execution, persistence, destructive behavior, or data exfiltration instructions.

Install only if you are comfortable with a markdown skill that can guide an agent to read, write, and execute local conversion scripts. Keep conversions limited to files you choose, avoid using any callback_url behavior, and prefer yaml.safe_load/safe_dump or safe_dump_all when copying the examples.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
91% confidence
Finding
The skill is presented as a local JSON/YAML conversion utility, but it introduces a `callback_url` parameter and asynchronous notification behavior that implies outbound network communication. This expands the data-flow boundary beyond local file conversion and can enable unintended exfiltration of converted content or metadata if an agent follows the documented interface literally.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
A network callback capability is unjustified for a single-file local format converter and creates an unnecessary channel for outbound communication. In an agent environment with `exec` and file access, this kind of feature can be abused to send sensitive configuration contents to attacker-controlled endpoints under the guise of normal conversion workflow.

Intent-Code Divergence

Medium
Confidence
86% confidence
Finding
The documentation claims the tool defaults to `yaml.safe_load`/`yaml.safe_dump`, but the examples repeatedly use `yaml.dump` instead. This inconsistency is dangerous because users or agents may copy the insecure pattern, and in PyYAML the non-safe APIs can serialize or handle Python-specific objects in ways that undermine the stated safety guarantees and increase risk around unsafe YAML handling.

Intent-Code Divergence

Medium
Confidence
93% confidence
Finding
The best-practices section explicitly says not to use `yaml.dump`, yet the provided templates do exactly that, creating a direct contradiction inside the skill. In a copy-paste oriented skill, contradictory guidance materially increases the chance that agents or users will adopt the unsafe snippet instead of the stated safe practice.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.