Security audit
Security checks for vulnerabilities and agentic risk
This is mostly a web-font helper, but it asks for broad write, command execution, API, and automation authority that is not well scoped to fonts.
Install only if you are comfortable granting this skill read, write, and command-execution capability in your agent environment. Prefer using it in a constrained workspace and treat any command execution, file writes, or API-key use as requiring explicit user confirmation.
The skill is presented as a narrowly scoped typography helper, but its documentation expands into generic automation capabilities. This kind of scope inflation can cause an agent or user to grant powerful file, execution, and integration permissions under false pretenses, increasing the chance of unsafe actions outside the expected typography domain.
Ambiguous, broad scope statements are dangerous here because the skill also declares powerful tools and external interactions. When a specialized skill describes itself as a general-purpose efficiency/automation component, it becomes easier to justify actions beyond the expected trust boundary.
The statement claiming risky code and external dependencies were removed conflicts with the continued presence of exec, write, API key setup, and external-service behavior. Misleading safety assurances are a red flag because they can reduce scrutiny and cause reviewers to trust a skill that still retains sensitive capabilities.
Overly broad applicability and invocation language can make the skill trigger in contexts far beyond typography. In agent ecosystems, ambiguous scope increases the probability that high-privilege tools are invoked for tasks the user did not intend this skill to handle.
Advertising command execution in a font-selection skill is risky because it normalizes use of powerful system capabilities unrelated to the core purpose. In an agent setting, this can encourage unnecessary shell access and make it easier for prompt-driven misuse or lateral actions to occur.
Mentioning file writing, command execution, and external API use without an up-front warning deprives users of informed consent about potentially state-changing or data-exposing behavior. In an agent environment, this can lead to unexpected writes, shell commands, or network transmission under a benign-looking skill label.
The skill advertises broad file processing, API integration, and information retrieval features that are not necessary for a typography advisor. This mismatch widens the effective attack surface and can mislead operators into approving unnecessary data access and outbound communication.
The skill metadata and documentation are partially fixed in Chinese, with mixed-language summaries, but do not state that users can choose their preferred output language or locale. This may violate a language/locale policy if the organization expects user choice rather than a default forced language experience.
No suspicious patterns detected.
