Security audit
Security checks for vulnerabilities and agentic risk
This is a simple Socratic-questioning skill with no code execution, credential access, persistence, or hidden data handling.
Install this if you want the agent to guide users through decisions by asking questions. Be aware it may activate for broad phrases like 'help me think it through,' and in factual, urgent, medical, legal, or crisis contexts the agent should leave Socratic mode and give direct or professional-guidance-oriented help.
The trigger phrases are broad, conversational, and likely to appear in ordinary user requests that do not necessarily require strict Socratic-only behavior. This can cause the skill to activate in unintended contexts, leading the agent to withhold direct answers or useful safety guidance when the user actually needs clear assistance.
The manifest descriptions on L06-L08 state that the skill responds with questions only and never with solutions. However, later documentation instructs the skill to 'directly give the answer' when the user asks for it or when the query is factual, which is an active contradiction in the skill's own documented intent.
The activation description includes phrases like "帮我捋一捋" and "help me think it through," which are common conversational requests and not uniquely tied to Socratic questioning. Although examples are provided, the trigger scope still overlaps with ordinary assistance requests, increasing the chance this skill is invoked when the user actually wants direct help or advice.
L31-L37 establishes a strict principle of asking without answering, advising, or hinting. But L84-L87 instruct the skill to offer options and, in some cases, directly provide an answer and rationale, which materially contradicts the earlier operational rules rather than merely omitting detail.
All user-facing documentation and trigger phrases are presented only in Chinese, which can amount to forcing a specific language without user opt-in. The file does not mention that the skill is intentionally Chinese-only or provide an option for other languages.
The file presents the operational instructions entirely in Chinese and does not state that the skill should match the user's preferred language. This creates a natural-language policy concern because the skill behavior appears locale-specific without documenting user opt-in or offering language flexibility.
The line stating the skill applies when a user is hesitant or struggling with a decision is broad and does not clearly distinguish between wanting reflective questioning and wanting concrete advice. This ambiguity could lead to the skill activating in many routine decision-support conversations where its question-only behavior is not desired.
No suspicious patterns detected.
