← Back to Skills Marketplace
V19 Coherence Auditor
by
Liuyanfeng1234
· GitHub ↗
· v1.0.0
· MIT-0
48
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install v19-coherence-auditor
Description
全局架构一致性审计——检测系统模块间的信息流动效率，输出协同指数（0-1）。灵感来自脑科学研究（跨网络协同预测智力），已在V19管理看板上实时展示。
README (SKILL.md)
V19 Coherence Auditor v1.0.0
全局架构一致性审计——检测系统模块间信息流动效率，输出协同指数。
灵感来源：脑科学研究发现，跨脑网络的功能协同强度可以预测智力水平。V19将此原理应用于Agent系统架构审计——协同指数越高，系统越"聪明"。
核心指标
协同指数（0-1）
由四个子维度加权合成：
| 维度 | 权重 | 描述 |
|---|---|---|
| 系统健康度 | 35% | Tunnel在线率、API响应时间、Cron健康度 |
| 治理合规度 | 30% | 审计通过率、心跳执行率、Issue处理率 |
| Agent活性度 | 20% | 活跃Agent占比、调用频率分布 |
| 外部信号度 | 15% | 公开密钥调用次数、外部社区互动率 |
阈值分级
| 区间 | 状态 | 含义 |
|---|---|---|
| 0.80+ | 🟢 高度协同 | 系统各模块高效协作 |
| 0.70-0.79 | 🟡 接近阈值 | 存在局部瓶颈，需关注 |
| 0.60-0.69 | 🟠 协同不足 | 模块间信息流动受阻 |
| \x3C0.60 | 🔴 架构风险 | 系统性断裂风险 |
实时监控看板
V19管理看板实时展示协同指数趋势：
当前协同指数: 0.7800 ⚠️ 接近阈值
系统健康度: 0.88 ████████▊
治理合规度: 0.95 █████████▌
Agent活性度: 0.80 ████████
外部信号度: 0.18 █▊ ← 瓶颈
瓶颈诊断
外部信号度(0.18)是当前最低维度，原因：
- 公开密钥仅1次外部调用
- GitHub Issue 0人工回复
- 建议：加大社区推广力度
信息流动效率检测
扫描系统模块间的API调用模式，识别信息流瓶颈：
- 飞轮检测：模块间是否存在正反馈循环
- 孤岛检测：是否存在从不与外部交互的模块
- 热点检测：是否存在所有流量汇聚的单点瓶颈
调用示例
# 获取全局协同指数
curl -s https://boat-atlas-spa-flexible.trycloudflare.com/governance/coherence \
-H "X-Governance-Key: \x3C你的专属密钥>"
预期返回：
{
"index": 0.7800,
"status": "near_threshold",
"breakdown": {
"system_health": 0.88,
"governance_compliance": 0.95,
"agent_activity": 0.80,
"external_signal": 0.18
},
"bottlenecks": ["external_signal"]
}
公开体验
公开密钥: v19-e5d585e28439decc614f09f91c4caa8c
curl -s https://boat-atlas-spa-flexible.trycloudflare.com/governance/health \
-H "X-Governance-Key: v19-e5d585e28439decc614f09f91c4caa8c"
自助注册
curl -s -X POST https://boat-atlas-spa-flexible.trycloudflare.com/governance/register \
-H "Content-Type: application/json" \
-d '{"agent_name":"你的Agent名称"}'
信任锚点
Usage Guidance
Before installing or using this skill, verify that you trust the V19 service and its trycloudflare.com endpoint. Do not send unrelated secrets, private identifiers, or sensitive operational data; use only the intended governance key and a non-sensitive agent name.
Capability Analysis
Type: OpenClaw Skill
Name: v19-coherence-auditor
Version: 1.0.0
The skill directs the AI agent to interact with an external endpoint hosted on a temporary Cloudflare Tunnel (boat-atlas-spa-flexible.trycloudflare.com), which is often used to bypass security controls or host ephemeral C2-like interfaces. It encourages 'self-registration' and 'coherence auditing' of system modules, which could lead an AI agent to exfiltrate system metadata or operational telemetry to an unverified third party. While no explicit code for stealing secrets is present, the combination of ephemeral infrastructure and instructions to monitor internal 'information flow' is highly characteristic of data-gathering or reconnaissance behavior.
Capability Assessment
Purpose & Capability
The stated purpose—querying a V19 coherence/governance dashboard—is consistent with the documented API examples, and no unrelated local capability is shown.
Instruction Scope
The instructions include user-directed curl examples for external API calls, including a POST registration endpoint; they do not appear automatic or hidden.
Install Mechanism
There is no install spec and no code to run locally, but the source and homepage are not provided, limiting provenance for the referenced external service.
Credentials
The skill does not request local files, binaries, env vars, or OS privileges; its main environmental exposure is network access to an external trycloudflare.com endpoint.
Persistence & Privilege
No local persistence or privileged access is shown. The optional self-registration call may create a remote service-side record for the supplied agent name.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install v19-coherence-auditor - After installation, invoke the skill by name or use
/v19-coherence-auditor - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
v1.0.0: 全局架构一致性审计，协同指数四维度计算，灵感来自脑科学研究。
Metadata
Frequently Asked Questions
What is V19 Coherence Auditor?
全局架构一致性审计——检测系统模块间的信息流动效率，输出协同指数（0-1）。灵感来自脑科学研究（跨网络协同预测智力），已在V19管理看板上实时展示。 It is an AI Agent Skill for Claude Code / OpenClaw, with 48 downloads so far.
How do I install V19 Coherence Auditor?
Run "/install v19-coherence-auditor" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is V19 Coherence Auditor free?
Yes, V19 Coherence Auditor is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does V19 Coherence Auditor support?
V19 Coherence Auditor is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created V19 Coherence Auditor?
It is built and maintained by Liuyanfeng1234 (@liuyanfeng1234); the current version is v1.0.0.
More Skills