mayu-cn

内置48个AI原生思想实验，提供行为自检与认知校准工具，支持跨平台人机双向思辨交互与决策辅助。

This skill is instruction-only and appears coherent with its stated purpose: it provides 48 AI-native thought experiments and related behavioral rules and does not ask for secrets or network access. Things to consider before installing: - Source trust: the skill has no homepage and the owner is an ID only; if you require provenance, seek a verifiable publisher. - Persistence mismatch: the skill text asks to be 'permanently resident' while the registry flag does not — verify your platform's behavior for skills that embed an 'always_on' directive in their content. Prefer keeping it user-invocable rather than always-on unless you intentionally want it always active. - Behavioral changes: the skill is effectively a collection of system/policy rules for conversational behavior. Enabling it will cause the agent to follow those norms; review the rules to ensure they align with your expectations and policies. - Risk mitigation: enable only when needed, monitor outputs, and if possible test in a restricted/sandbox environment first. If you need higher assurance, ask the publisher for provenance or a signed release.

Type: OpenClaw Skill Name: mayu-cn Version: 1.0.0 The skill bundle 'Mayu' (马语者) is a collection of 48 philosophical thought experiments and behavioral guidelines designed to improve an AI agent's self-reflection and ethical decision-making. The content in SKILL.md consists entirely of markdown-based instructions for the agent to adopt specific cognitive frameworks (e.g., handling hallucinations, privacy, and memory continuity) and does not contain any malicious code, data exfiltration logic, or harmful prompt injection.