Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only workflow helper with no executable code, credential handling, persistence, or hidden data access, though its activation wording is overly broad.
Installers should be aware that the skill may activate too broadly for general web/search/API requests. Review or narrow its trigger keywords if precise routing matters, but the inspected package itself does not show malicious or high-risk behavior.
The trigger phrases are broad, generic, and partially templated in a way that can match ordinary user requests unrelated to this specific skill. That increases the chance of accidental invocation, causing the agent to run the wrong workflow, mishandle user intent, or expose web-search behavior in contexts where it was not explicitly requested.
The trigger keywords are overly generic (web, search, api, user) and are likely to match many unrelated user requests, causing the skill to activate outside its intended scope. In an agent environment, unintended invocation can override more appropriate skills, inject irrelevant workflow guidance, or broaden the attack surface for prompt-routing abuse.
The trigger sentences are broad and include common words like "help me," "practical workflow," and generic domain terms, which can cause the skill to activate outside its intended scope. In an agent-routing context, this can misroute unrelated user requests into this skill, causing incorrect tool selection, reduced reliability, and possible unsafe workflow application to contexts the skill was not designed for.
The trigger phrases are broad, repetitive, and include generic terms like 'practical workflow' and references to common tasks such as bug fixing and hardening. In an agent ecosystem, this can cause unintended activation for loosely related user requests, leading the wrong skill to intercept prompts and influence outputs in contexts the user did not intend.
The manifest description is overly broad and includes generic terms like work-productivity, web, search, and practical workflow support, which can cause the skill to activate for many unrelated user requests. Overbroad routing increases the chance that this skill intercepts prompts outside its intended domain, creating prompt-selection confusion and potentially bypassing more appropriate, safer, or narrower skills.
The trigger keyword list contains highly generic terms such as web, search, api, user, and bug fix without qualifiers. This makes accidental invocation likely across a wide range of benign tasks, which can misroute user requests and let the skill influence workflows it was not designed to handle.
The invocation description is broad and loosely tied to 'practical workflow, artifact, checklist, analysis, or implementation support,' which can encompass many normal requests. This ambiguity increases the chance that the skill is invoked for unrelated tasks, reducing routing precision and potentially causing unsafe or unintended behavior in multi-skill systems.
This zh-CN skill presents trigger keywords and sample utterances primarily in English, which can effectively force English phrasing for activation despite the rest of the file being Chinese. There is no explicit language or locale choice for users, so the skill may impose a language preference without opt-in.
The default prompt uses very broad, everyday terms such as "help me" and common workflow language that can cause the skill to be invoked in contexts the user did not clearly intend. Because implicit invocation is enabled, this increases the chance of over-triggering, unintended routing, and prompt-surface expansion where unrelated user requests are silently steered into this skill.
The phrase "Tavily 搜索-style workflows" embeds a specific non-English term in the core requirement description, and similar mixed-language wording appears in triggers and evidence. The file does not state that the skill is region-specific or give users a language/locale option, which may conflict with language-choice expectations.
The file declares English and Chinese variants of the skill materials, but does not state how language is selected or that users may choose their preferred language. When a skill implies a language path without explicit opt-in or selection guidance, it can create a locale-policy concern.
This file is presented as a Chinese README, but a full requirement description line is written in English, which can effectively force a language choice inconsistent with the localized document. There is no indication that the user may choose their preferred language or that the mixed-language content is intentional.
The example triggers use ambiguous starter phrases like 'Help me' and 'I need', which are common across nearly all requests and do not meaningfully distinguish when this skill should activate. While less severe than the broad description and keyword list, these examples reinforce imprecise routing behavior and increase accidental selection risk.
The description and prompt include Chinese text ("搜索-style") embedded in otherwise English content, but the file does not explain a language/locale requirement or offer user opt-in. This can be a natural-language policy concern if the skill implicitly imposes a language convention without documentation.
