Security audit
Security checks for vulnerabilities and agentic risk
This is a document-formatting guidance skill with broad triggers but no hidden code, credential access, persistence, or automatic destructive behavior.
Installers should treat this as a lightweight Word/DOCX workflow helper. Use copies or backups before applying any bulk formatting or find/replace advice to important documents, and be aware that its broad implicit trigger settings may cause it to appear for loosely related productivity prompts.
The trigger phrases are generic and repetitive enough that the skill could activate for a wide range of unrelated productivity requests. Over-broad activation increases the chance of incorrect routing, unintended invocation, and interference with other skills or user workflows, though it does not directly create code-execution risk in this README alone.
The README does not warn users that formatting repair, find/replace, and style normalization can make bulk or irreversible document changes. Without user-facing cautions, users may apply suggested workflows directly to important files, creating integrity and availability risks such as corrupted layout, lost formatting, or accidental edits in legal or operational documents.
The trigger phrases are generic and broad enough that the skill could activate for loosely related requests, increasing the chance of unintended invocation. In a document-editing skill, accidental activation can lead to inappropriate formatting guidance or modification workflows being applied to the wrong task or sensitive documents.
Most user-facing headings and usage guidance in this README are presented in Chinese, but the document does not state that language selection is optional or user-chosen. For organizational language/locale policy, presenting a locale-specific experience without opt-in can be a natural-language policy violation unless clearly justified as region-specific.
The description allows invocation for very broad categories like work-productivity, document formatting, styles, and implementation support, which can match many ordinary requests outside a narrowly scoped Word DOCX repair use case. Over-broad routing can cause the agent to activate this skill in unintended contexts, increasing the chance of inappropriate instructions, prompt-surface expansion, or interference with more suitable skills.
The trigger keywords are generic terms that commonly appear in harmless everyday requests and are not sufficiently constrained to the skill's intended scope. This makes accidental or excessive activation more likely, which can misroute user prompts and broaden the attack surface for prompt injection or inappropriate capability selection.
The example trigger sentences use everyday phrasing like 'Help me' and 'I need a practical workflow' combined with truncated broad requirement text, which provides little routing specificity. In practice, such examples encourage invocation on loosely related requests and can cause the skill to capture prompts it was not designed to handle.
该技能文件整体以 zh-CN 语言环境提供，但未见向用户提供语言选择、切换机制，或说明该语言限制是区域合规/业务所必需。根据 SQP-3，强制特定语言而无用户选择可能构成自然语言层面的组织政策问题。
The skill description defines a very wide invocation scope covering broad categories like work-productivity and Microsoft Word without meaningful boundaries. This can lead to over-selection of the skill, causing context confusion, incorrect task framing, and accidental handling of requests that should be routed to more specific or safer skills.
The trigger keywords are broad and overlap with common productivity conversations, increasing the chance that this skill is invoked when the user did not intend document-formatting assistance. Misrouting can cause the agent to apply the wrong workflow, assumptions, or automation guidance to unrelated tasks, reducing reliability and potentially causing unsafe or irrelevant actions in downstream tool-enabled contexts.
The default_prompt uses a very broad, natural-language invocation phrase tied to common productivity tasks like document formatting, find-and-replace, and styles. Because the skill also allows implicit invocation, this broad wording can cause the agent to activate in many ordinary conversations where the user did not explicitly request this skill, increasing the chance of unintended tool use and prompt-scope expansion.
Enabling allow_implicit_invocation without strong activation constraints creates a routing vulnerability: the system may invoke this skill automatically based on weak semantic matches. In a broad office-productivity context, that can lead to unauthorized or unnecessary invocation, accidental disclosure of user context to the skill, or interference with normal assistant behavior.
The evidence and linked titles are presented entirely in Chinese, with no indication that language choice is optional or justified by a region-specific requirement. This can create a language/locale policy issue if the skill implicitly assumes Chinese-language content without offering user choice.
The invocation description and trigger block lack specificity about when the skill should and should not activate, making it easy for normal conversational text to match. This increases the risk of accidental invocation, inappropriate context capture, and misapplication of the skill in situations where the user did not intend document-formatting assistance.
The trigger sentence is extremely broad and can match ordinary user phrasing unrelated to a deliberate request for this specific skill. Overbroad activation can cause unintended skill invocation, which may route users into the wrong workflow, suppress better-matched skills, or let a weakly scoped skill influence outputs without clear user intent.
No suspicious patterns detected.
