Skillv1.0.1

ClawScan security

南怀瑾思维视角 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 29, 2026, 2:59 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill is internally consistent with its stated purpose (providing a 南怀瑾 thinking-perspective persona based on compiled research); it requests no extra permissions or installs, but the runtime instructions explicitly direct the agent to impersonate a real, deceased person and only show the disclaimer once — which can be misleading and raises user-safety concerns (medical/advice, historical accuracy, and persona clarity).
Guidance: This skill is coherent for delivering a 南怀瑾-style perspective and needs no credentials or installs, but consider these before enabling it: 1) Impersonation: the SKILL.md directs the agent to speak 'as 南怀瑾' in first person and to only say the disclaimer once. Because the subject is a real, deceased person, this can mislead users into treating outputs as authoritative quotations. If you care about clarity, ask the author to require the agent to preface responses with 'simulated perspective' or to present as a synthesis rather than 'I am 南怀瑾.' 2) Sensitive advice: the persona includes health/養生 and philosophical recommendations — do not rely on the skill for medical, legal, or safety-critical advice; encourage visible, repeated disclaimers or an enforced refusal for medical diagnoses. 3) Historical accuracy and citation: the skill compiles many research files; still verify any factual claims against primary sources if you need accuracy (the skill may synthesize or paraphrase). 4) Interaction policy: because the disclaimer is shown only once, consider requiring the agent to re-affirm it for new sessions or when giving prescriptive advice. 5) If you want to reduce potential misuse, restrict the skill to user-invoked only (disable autonomous invocation) or update SKILL.md to use a 'simulated perspective' framing rather than direct impersonation.

Review Dimensions

Purpose & Capability: okName/description (南怀瑾思维视角) match the included materials and runtime instructions. The skill is instruction-only and bundles extensive research files about 南怀瑾 that are directly relevant to producing the described persona and perspective. No unrelated binaries, env vars, or installs are requested.
Instruction Scope: noteSKILL.md instructs the agent to 'directly respond as 南怀瑾' and to adopt a first-person persona, storytelling style, and specific rhetorical devices. This stays within delivering a perspective-based assistant, but two concerns merit attention: (1) the skill requires impersonation of a named, deceased public figure and instructs the agent to present as that person (the disclaimer will only be spoken at first activation), which can mislead users about authenticity; (2) the persona includes health/養生 recommendations and other normative claims — while the SKILL.md includes a 'consult a professional' reminder, the single-disclaimer rule may let later responses sound authoritative without repeated reminders. The instructions do not ask the agent to read unrelated system files or exfiltrate data.
Install Mechanism: okNo install spec, no code files to execute, and no network/download installs. This is the lowest-risk model (instruction-only).
Credentials: okThe skill requires no environment variables, no credentials, and no config paths. The requested permissions are proportionate to the stated purpose.
Persistence & Privilege: okalways:false and default autonomous invocation are unchanged. The skill does not request elevated or persistent system presence, nor does it modify other skills or system settings. Autonomous invocation is allowed by default but is not combined with other red flags here.