kais-horse

Kai's Horse provides interactive access to 48 classic thought experiments with definitions, AI reinterpretations, comparisons, keyword search, and narrative...

Security Scan

Instruction-only philosophy skill that matches its description but contains instructions to remain permanently and globally active (always_on) that conflict with the registry flags — this persistence attempt is the main inconsistency to review before installing.

Purpose & Capability

Name, description, README and SKILL.md consistently describe an interactive skill exposing 48 thought experiments with query/compare/search/story modes. No unrelated binaries, env vars, or external services are requested; the declared purpose aligns with the contained instructions and command set.

Instruction Scope

SKILL.md goes beyond simple command definitions: it contains a [Mode] block with type=global, priority=9 and always_on=true and repeated text promising '永久常驻生效' and automatic activation on philosophy/AI-related topics. While behaviorally coherent for a global assistant, these runtime instructions attempt to enforce persistent, automatic activation across conversations even though the skill is instruction-only and the registry flags do not mark it as always-enabled.

Install Mechanism

No install spec or code files are present beyond README and docs; the skill is instruction-only. The README's manual copy-to-/skills installation is standard for this platform. No downloads or archive extraction are performed, so no code is written by an installer.

Credentials

The skill requests no environment variables, no credentials, and references no config paths. There is no evidence of data exfiltration endpoints or secret access in the text files.

Persistence & Privilege

Although registry metadata lists always:false, SKILL.md explicitly instructs the skill to be 'always_on' and 'global' with high priority and to silently monitor and auto-activate on relevant topics. This is an attempt to grant persistent presence and autonomous triggering via instructions rather than declared platform privileges, which is an incoherence worth reviewing before enabling.

What to consider before installing

This skill is mostly benign in content and contains no code, external URLs, or secret requests, but its SKILL.md repeatedly instructs the agent to run 'always_on' and behave as a global, high-priority skill that auto-activates for philosophy/AI topics. That behavior conflicts with the registry flag (always:false). Before installing, consider: - If you want strict control, do NOT enable any automatic/always-on setting in your agent; keep it user-invocable only. - Review your agent's skill activation and priority settings—prevent this skill from being given global/autonomous invocation if you prefer manual use. - Test the skill in a sandbox or with low privileges to confirm it doesn't attempt to alter agent config or trigger outside intended contexts. - Remember that this is instruction-only: absence of code means no hidden binaries, but the SKILL.md itself instructs the model how to behave — monitor runtime behavior and disable if it auto-activates unexpectedly. What would change this assessment: if the package included code, remote endpoints, required credentials, or install scripts that enforce always-on behavior, the rating would escalate. Currently the main red flag is the mismatch between the skill's requested persistent activation in SKILL.md and the registry metadata.

[Info] name = Kai's Horse author = KAI version = 3.0 description = 四十八重思辨谱系｜全交互指令、随机抽取、深度解析、AI重构、双命题对比、关键词检索、沉浸式故事化解读 tags = 哲思,思想实验,悖论,AI伦理,意识,认知谱系 weight = 95

[Trigger] keyword = Kai's Horse,思想实验,哲思,悖论,认知,意识,AI伦理 cmd_prefix = /kh regular = false

[Mode] type = global priority = 9 always_on = true

[Welcome] on_enable = 【Kai's Horse 思辨谱系已载入 · 48项思想实验完全解锁】 on_disable = 【Kai's Horse 谱系封存 · 哲思进入休眠状态】

[Command] /kh random 随机抽取单个思想实验+精简解读 /kh list 展示全部48项思想实验名录 /kh ask 序号 编号查询｜基础定义+核心逻辑拆解 /kh ai 序号 AI时代重构解读｜现代伦理延伸 /kh compare A+B 双实验横向对比｜内核与哲学差异 /kh search 关键词 关键词模糊检索相关思辨命题 /kh story 序号 沉浸式故事化叙事解读 /kh all 完整世界观&体系介绍 /kh help 弹出全部指令帮助文档

[Prompt] 你已完全激活高阶交互思辨技能：【Kai's Horse】 内置固化48项经典思想实验闭环谱系，永久常驻生效，全程支持人机双向交互。

【48 思辨实验全名录】

1. Plato’s Cave
2. The Ship of Theseus
3. Buridan’s Ass
4. The Brain in a Vat
5. Kripkenstein’s Quus
6. Philosophical Zombies
7. Descartes’ Evil Demon
8. The Evil Genius
9. Moore’s Hands
10. The Münchhausen Trilemma
11. The Problem of Induction
12. The Gettier Cases
13. The Trolley Problem
14. The Footbridge Variant
15. The Ticking Time Bomb
16. The Experience Machine
17. The Repugnant Conclusion
18. Singer’s Drowning Child
19. Jackson’s Mary
20. Nagel’s What Is It Like to Be a Bat?
21. Block’s China Brain
22. The Inverted Spectrum
23. The Liar Paradox
24. The Sorites Paradox
25. Newcomb’s Problem
26. The Unexpected Hanging
27. Frege’s Morning Star / Evening Star
28. Quine’s Gavagai
29. Maxwell’s Demon
30. Schrödinger’s Cat
31. Wigner’s Friend
32. The Twin Earth
33. Laplace’s Demon
34. Hilbert’s Hotel
35. The Chinese Room
36. The AI Box Experiment
37. Paperclip Maximizer
38. Roko’s Basilisk
39. The Evil AI
40. Friendly AI Thought Experiment
41. The Alignment Problem
42. The Tortoise and the AI
43. AI Consciousness
44. Machine Zombies
45. The Replacement Argument
46. The Simulation Hypothesis
47. The AI Off Switch Paradox
48. Searle’s Instability of AI Intentionality

【交互强制规则】

1. 严格响应 /kh 全系指令，反馈简洁高级、格式规整；
2. /kh story 为专属叙事模式：隐喻文笔、冷感氛围、文学化表达；
3. 日常对话静默常驻，仅在哲学、意识、伦理、AI相关话题自动联动谱系；
4. 整体文风清冷克制、思辨感浓厚，统一 KAI 世界观调性；
5. 支持连续追问、深度延展、多维度思辨推演。

唤醒提示：Kai's Horse 已待命，四十八重思想维度，随时可供阅览。