Security audit
Security checks for vulnerabilities and agentic risk
This JSON linting skill is not clearly malicious, but its instructions broaden a local validator into unrelated file, API, network, and command-execution use cases.
Review this skill before installing. It appears to be a local JSON validator with no embedded code, but its instructions are overbroad enough that an agent could invoke read and exec capabilities for tasks beyond JSON linting. Only use it if you are comfortable constraining it to explicit .json validation commands in a known workspace.
The manifest description includes broad trigger phrases such as file processing, document conversion, format conversion, and content extraction, which go well beyond JSON syntax checking. Overbroad invocation cues are dangerous because they can cause the agent to select this skill in unrelated situations, exposing exec/read privileges where a narrower or safer tool should have been used.
The capability coverage section enumerates vague and expansive scenario keywords, including unrelated analytics/reporting language, which can drive unintended invocation. In an agent ecosystem, this acts like prompt-surface inflation: the skill becomes eligible for many tasks outside its legitimate purpose while still retaining command execution capability.
The manifest presents a narrow local JSON syntax checker, but the body expands the scope to generic file handling, external API integration, command execution, and information retrieval. Scope expansion like this is dangerous because agents may invoke the skill in contexts far beyond linting, causing unjustified file or command access under the cover of a benign-seeming tool.
The description explicitly states '支持中文交互', and the document consistently frames operation in Chinese without offering a language choice. This can violate language/locale policy when a skill defaults to a specific language rather than letting the user opt in or select their preferred language.
The capability description at L61 and best-practices section at L176-L180 describe historical trend comparison and tracking, implying this skill supports persisted history analysis. But the limitations section at L208 states historical trend tracking and monitoring alerts are restricted in the free version. This is an active contradiction about what the current skill version does.
The skill claims all validation is local and requires no external services, yet later introduces network testing guidance and generic external API integration language. This kind of contradictory documentation can mislead an agent or operator into granting broader execution or network behavior than the skill’s stated purpose justifies, weakening trust boundaries around a supposedly local-only tool.
Claiming external API integration in a local JSON validator is unjustified capability inflation. Even without an explicit network tool in the manifest, such language can normalize or prompt broader behavior from an agent, increasing the chance of over-privileged execution or unexpected data flow beyond the workspace.
No suspicious patterns detected.
