Skillv1.0.0

ClawScan security

文言 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignApr 29, 2026, 5:28 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: This is a purely stylistic instruction-only skill that tells the agent to answer in classical Chinese; its requirements and instructions are proportional to that purpose and it asks for no secrets or installs.
Guidance: This skill is stylistic only: it tells the agent to reply in classical Chinese and includes guidance for exceptions (code, machine-readable content, plain-language explanations). It asks for no installations or credentials. If you want to avoid automatic stylistic changes, you can disable or not invoke the skill; otherwise you can safely use it knowing it only affects output phrasing and persists the selected register within the conversation. If you need assurance the agent won't make external network calls, note this skill does not include any code that would perform such actions, but the underlying agent platform could still contact external services per its own configuration.

Purpose & Capability: okName/description (force replies into wenyanwen) match the skill contents: the SKILL.md and style reference define registers and examples. The skill asks for no binaries, env vars, or external credentials that would be unrelated to a formatting/style layer.
Instruction Scope: okSKILL.md confines behavior to changing natural-language output style, preserving machine-readable text unchanged, and obeying explicit exception rules. It does not instruct reading files, environment variables, or contacting external endpoints. It does request persisting the selected register for the conversation, which is a normal conversational-state instruction.
Install Mechanism: okNo install spec and no code files that perform downloads or writes. Instruction-only skills are lowest risk and nothing is written to disk by the skill itself.
Credentials: okThe skill requires no environment variables, credentials, or config paths. There are no secrets requested and no broad access requested.
Persistence & Privilege: noteThe skill requests that the agent 'persist the selected register for the rest of the conversation' (conversation-level state), and agents/openai.yaml sets allow_implicit_invocation: true. Autonomous invocation is the platform default and not itself a concern, but be aware the style can be reapplied automatically during the conversation until the user changes it.