Security audit

JSON校验工具免费版

Security checks across malware telemetry and agentic risk

Overview

This JSON lint skill is mostly a local checker, but its instructions include unrelated broad file-processing, API, network, and command-execution language that makes its real scope unclear.

Install only if you want a local JSON syntax-checking helper and are comfortable with it reading workspace .json files and using command execution for parsing. Prefer running it on a chosen directory with exclusions such as node_modules, .git, dist, and sensitive data folders; do not use its broader API, network, conversion, or content-extraction language as authorization for unrelated tasks.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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

Findings (6)

Description-Behavior Mismatch

Medium

Confidence: 92% confidence
Finding: The skill is presented as a local JSON syntax checker, but the later 'core functions' and related sections expand its scope to generic file handling, API integration, command execution, and information retrieval. This scope drift weakens user consent and can cause an agent to invoke broader capabilities than expected, especially since the skill is allowed to use exec.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: Introducing network and external-service behavior in a tool that should operate entirely locally creates unnecessary expansion of attack surface and user expectations mismatch. Even without explicit network tools in the manifest, instructing agents to perform connectivity checks or external interactions can lead to unsafe workflow composition in environments where other tools are available.

Intent-Code Divergence

Medium

Confidence: 91% confidence
Finding: The documentation claims no extra API key is needed and all validation is local, yet elsewhere it discusses API-key handling and external API integration. These contradictory statements make the skill's trust boundary unclear and can mislead users or agents into exposing secrets or enabling unnecessary external access.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The top-level description uses broad invocation language such as file processing, document conversion, format conversion, and content extraction, which materially exceeds JSON linting. Overbroad routing language can cause the agent to select this skill in unrelated contexts, exposing workspace files to recursive scanning and exec-backed actions the user did not intend.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The capability coverage section contains expansive and ambiguous keywords that include unrelated analysis, reporting, visualization, and workflow scenarios. This increases the chance of unintended activation and scope creep, which is risky for a skill with filesystem scanning and command execution permissions.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The skill recursively scans workspace .json files and reports file paths plus parse-derived error details, but it does not prominently warn users about this visibility. In sensitive repositories, path names and error snippets can disclose confidential project structure or data-derived information beyond what the user expects from a simple linting request.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.