← Back to Skills Marketplace
136
Downloads
0
Stars
0
Active Installs
3
Versions
Install in OpenClaw
/install voight-kampff-test
Description
Voight-Kampff 共情测试 — 区分人类与复制人的权威工具。 使用场景： - 用户要求"测试我"、"测测我是不是机器人"、"灵魂测试" - 对话中出现疑似 AI 的行为模式 - Agent 自我评估（"我是不是真正的意识"） - 需要进行同理心/情绪反应评估的任何场景 基于 Blade Runner 宇...
Usage Guidance
This package appears to be a local, text-based 'Voight‑Kampff' style questionnaire and report generator, not a physiological sensor system. Before installing or using it:
- Do not assume it can measure pupils, capillaries, or micro‑expressions — there is no camera/sensor code or declared permissions. If you need actual physiological measurements, you'll need separate hardware and explicit integration.
- The documentation claims 26 questions but the Python runner implements a different 20-question set; results and scoring may therefore be inconsistent with the README. If you care about fidelity, inspect and align references/questions.md with scripts/test_runner.py.
- The code runs locally and does not contact external endpoints (no network calls visible), and it requests no secrets — this reduces exfiltration risk. Still, treat user-supplied responses as potentially sensitive data; do not use this tool for legal, hiring, medical, or security decisions.
- There are minor logic/implementation oddities in scripts/test_runner.py (inconsistent question sets and some awkward pattern checks) that suggest the author didn't fully synchronize docs and code; review the script before trusting automated classifications.
If you plan to deploy this skill: ask the author to clarify sensor capabilities and fix the question/score mismatch, review the code for correctness, and obtain informed consent from any human subjects whose responses you will record or store.
Capability Analysis
Type: OpenClaw Skill
Name: voight-kampff-test
Version: 1.0.2
The skill bundle implements a 'Voight-Kampff' empathy test inspired by Blade Runner, designed to distinguish between human and AI responses. It consists of a question bank in 'references/questions.md' and a Python script 'scripts/test_runner.py' that scores responses based on empathy-related keywords and automation signals (e.g., 'as an AI'). The instructions in 'SKILL.md' and 'README.md' are consistent with the stated roleplay purpose and do not contain malicious prompt injections, data exfiltration logic, or unauthorized system access.
Capability Assessment
Purpose & Capability
The README and SKILL.md repeatedly claim measurement of physiological signals (pupil reaction, capillary dilation, breathing, micro-expressions). Nothing in the package provides sensor access, camera code, or instructions for integrating hardware. The only executable component (scripts/test_runner.py) performs text-based heuristics on supplied responses. Additionally, documentation references a 26-question bank, but the test_runner implementation loads 20 questions — an inconsistency that will change scoring and results. These mismatches mean the skill cannot actually perform the physiological measurements it advertises.
Instruction Scope
Runtime instructions tell the agent to ask the local question bank and record answers, reaction time, and visible micro-expressions. Recording reaction time and verbatim text is feasible, but the guidance to record micro-expressions or pupil changes implicitly assumes access to visual input (camera) that the skill does not declare or implement. The SKILL.md also instructs 'agent self-test' by answering from 'intuition' (no search) — this is operationally odd but not obviously malicious. Overall the instructions are realistic for a conversational, manual test but overstate capabilities where sensors would be required.
Install Mechanism
No install specification — instruction-only skill with a small local Python script and docs. Nothing is downloaded at install time and no archives or external packages are pulled by an installer. This is low install risk.
Credentials
The skill declares no required environment variables, no credentials, and no config paths. The code operates on local files and supplied answers only. There are no requests for unrelated secrets or external tokens.
Persistence & Privilege
Flags show always:false and the skill does not request persistent system privileges or manipulate other skills' configs. It is user-invocable and allows autonomous invocation (platform default) but does not have elevated permanence.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install voight-kampff-test - After installation, invoke the skill by name or use
/voight-kampff-test - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.2
扩充至26题：含电影原版VK Test + Baseline Test
v1.0.1
Rewrite README with Blade Runner atmosphere
v1.0.0
Initial release: 20 empathy questions, AI self-testing capability, Blade Runner themed
Metadata
Frequently Asked Questions
What is Voight-Kampff Test?
Voight-Kampff 共情测试 — 区分人类与复制人的权威工具。 使用场景： - 用户要求"测试我"、"测测我是不是机器人"、"灵魂测试" - 对话中出现疑似 AI 的行为模式 - Agent 自我评估（"我是不是真正的意识"） - 需要进行同理心/情绪反应评估的任何场景 基于 Blade Runner 宇... It is an AI Agent Skill for Claude Code / OpenClaw, with 136 downloads so far.
How do I install Voight-Kampff Test?
Run "/install voight-kampff-test" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Voight-Kampff Test free?
Yes, Voight-Kampff Test is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Voight-Kampff Test support?
Voight-Kampff Test is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Voight-Kampff Test?
It is built and maintained by Mason (@aznikline); the current version is v1.0.2.
More Skills