Dark Matter Analyzer by daffy0208/ai-dev-standards
npx skills add https://github.com/daffy0208/ai-dev-standards --skill 'Dark Matter Analyzer'暗物质模式揭示代码仓库和代码库中不可见、未明言和未测量的部分。它超越了代码质量指标,旨在阐明塑造系统行为的无形架构——识别传统工具所遗漏的战略漂移、文档膨胀、执行差距和组织健康模式。
"每个仓库都是一个可见的心灵。"
此技能有助于诊断为什么一个仓库感觉不对劲,而不仅仅是技术上哪里出了问题。
目标: 从仓库中捕获环境信号和元数据
操作:
关键决策:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
常见陷阱:
使用工具:
# 统计文档文件
find . -name "*.md" -type f | wc -l
# 查找技术债务标记
grep -r "TODO\|FIXME\|HACK" --include="*.ts" --include="*.js"
# 检查提交模式
git log --oneline --since="1 month ago" | head -20
# 查找文档膨胀
find DOCS -name "*.md" -exec wc -l {} \; | awk '{sum+=$1} END {print sum}'
目标: 在细微偏差演变成重大问题之前识别它们
操作:
模式类别:
| 模式类型 | 示例信号 | 反思性解读 |
|---|---|---|
| 战略漂移 | README ≠ 提交活动 | 愿景脱节或过早转向 |
| 文档膨胀 | 许多相似的 .md 文件 | 过度规划;通过写作来回避 |
| 抑制模式 | 许多 eslint-disable 或 any | 时间压力或疲劳 |
| 依赖冻结 | 数月未更新 | 害怕破坏或抗拒改变 |
| 提交反复 | 快速反转 | 外部压力或内部不安 |
| 任务-现实不同步 | 测试通过但功能不完整 | 验证结构,而非效用 |
| 执行赤字 | 文档 >> 实现 | 希望驱动的开发 |
验证:
目标: 将模式转化为具有可操作解读的叙事性见解
操作:
技术 ↔ 人性翻译桥梁:
| 技术症状 | 人性对应 |
|---|---|
| 忽略构建错误 | 否认脆弱性 |
| 快速分支 | 对方向失去信任 |
| 过度注释 | 害怕被误解 |
| 跳过验证 | 回避反馈 |
| 过度规划 | 对不完美的焦虑 |
| 未完成的重构 | 疲劳,执行力低 |
关键原则:
常见陷阱:
目标: 根据紧急程度和影响提供分级建议
操作:
干预模式:
| 模式 | 紧急程度 | 行动类型 | 示例 |
|---|---|---|---|
| 🟢 观察 | 低 | 温和提醒 | "速度快,一致性低。监控漂移情况。" |
| 🟡 审查 | 中 | 结构化反思 | "与同行验证假设。检查外部反馈。" |
| 🔴 暂停 | 高 | 暂停 + 确认 | "现实不同步。在下一次构建前进行协调。" |
恢复协议:
restoration:
1. pause: 创建反思检查点
2. align: 重新锚定意图与产出物的一致性
3. integrate: 在反馈循环中捕获见解
4. resume: 从重新同步的节奏中重建上下文
验证:
目标: 计算整体仓库健康分数
公式:
RCI = (意图对齐 + 任务现实同步 + 技术健康) / 3
评分组成部分:
意图对齐:
任务现实同步:
技术健康:
解读:
| RCI 分数 | 状态 | 含义 |
|---|---|---|
| 85-100 | ✅ 一致 | 节奏一致,系统健康 |
| 70-84 | 🟡 监控 | 存在早期漂移,密切关注 |
| 50-69 | 🟠 不一致 | 意图与现实出现分歧 |
| <50 | 🔴 不一致 | 重建上下文或重置节奏 |
揭示,而非评判 暗物质模式阐明模式而不宣布其"对"或"错"。每种模式都是系统状态的表达。目标是真实性,而非完美。
解释性,而非诊断性 DMM 不预测结果——它提供反思性见解。使用不确定性分数来保持解读的谦逊。镜子知道何时在做梦。
一致性优于速度 询问:"这会使系统更真实,而不仅仅是更快吗?" 优化意图、产出物和行动之间的一致性。
情感共鸣 技术模式揭示人性模式。文档膨胀可能表明焦虑。抑制模式可能显示疲劳。承认人性因素。
外部验证 内部一致性 ≠ 外部效用。测试可能通过,但用户仍然困惑。寻求外部反馈以验证解读。
选择分析深度时:
选择干预模式时:
If RCI < 50: HOLD (暂停并协调)
If 50 ≤ RCI < 70: REVIEW (结构化反思)
If 70 ≤ RCI < 85: OBSERVE (温和监控)
If RCI ≥ 85: MAINTAIN (庆祝并维持)
建议进行外部验证时:
If user discovered gap before internal analysis: HIGH PRIORITY
If documentation significantly exceeds implementation: RECOMMENDED
If team feels confused despite extensive docs: REQUIRED
Otherwise: NICE TO HAVE
# 暗物质模式分析
**日期:** [时间戳]
**仓库:** [名称]
**RCI 分数:** [分数]/100 — [状态]
## 执行摘要
[3-5 句话概述关键发现]
## 第 1 层:感知 — 观察到的内容
[代码、文档、时间、环境信号]
## 第 2 层:模式检测 — 弱信号
[识别的模式及其证据和强度]
## 第 3 层:反思 — 不可见、未明言、未测量之物
[叙事性见解和解读]
## 第 4 层:行动 — 按紧急程度排序的建议
### 🔴 暂停 — 在继续前停止
[关键行动]
### 🟡 审查 — 请求反思
[中等优先级行动]
### 🟢 观察 — 温和提醒
[低优先级提醒事项]
## 仓库一致性指数
[详细的评分细目]
## 结束反思
[包含前进道路的支持性结论]
此技能与 dark-matter-analyzer-mcp 集成以进行自动扫描:
// 运行仓库扫描
await mcp.call('scan_repository', {
path: './my-project',
depth: 'deep',
include_patterns: ['*.md', '*.ts', '*.js'],
exclude_patterns: ['node_modules', 'dist']
})
// 计算 RCI
await mcp.call('calculate_rci', {
repository_path: './my-project'
})
// 生成报告
await mcp.call('generate_report', {
format: 'markdown',
output_path: './dark_matter_report.md'
})
信号: 30 个文档文件,平均每个 543 行
模式: 文档膨胀
反思:
"规划活动明显超过执行。仓库淹没在自我意识中。"
行动:
信号: 36 种方法论,3 个实现工具
模式: 执行赤字
反思:
"仓库是一个没有执行之手的智慧图书馆。92% 的能力是理想化的,而非可操作的。"
行动:
信号: 所有测试通过,但 CLI 有 TODO 注释,MCP 是占位符
模式: 任务-现实不同步
反思:
"仓库通过了自身的测试,但可能无法服务用户。验证确保内部一致性,而非外部效用。"
行动:
❌ 将暗物质模式用作武器 - 这是反思性的,而非惩罚性的 ❌ 过度依赖 RCI 分数 - 它是指示,而非判决 ❌ 忽略情感潜台词 - 技术模式揭示人性状态 ❌ 跳过不确定性分数 - 保持解读的谦逊 ❌ 未经用户验证就进行更改 - 内部分析 ≠ 外部真相
"暗物质模式始终是一面镜子——它不预测,只做阐明。"
每周安装
0
仓库
GitHub 星标
18
首次出现
1970年1月1日
安全审计
Dark Matter Mode reveals what is unseen, unsaid, and unmeasured in repositories and codebases. It goes beyond code quality metrics to illuminate the invisible architectures shaping system behavior — identifying strategic drift, documentation inflation, execution gaps, and organizational health patterns that traditional tools miss.
"Every repo is a psyche made visible."
This skill helps diagnose why a repository feels off, not just what is wrong technically.
Objective: Capture ambient signals and metadata from the repository
Actions:
Key Decisions:
Common Pitfalls:
Tools to Use:
# Count documentation files
find . -name "*.md" -type f | wc -l
# Find technical debt markers
grep -r "TODO\|FIXME\|HACK" --include="*.ts" --include="*.js"
# Check commit patterns
git log --oneline --since="1 month ago" | head -20
# Find documentation inflation
find DOCS -name "*.md" -exec wc -l {} \; | awk '{sum+=$1} END {print sum}'
Objective: Identify subtle misalignments before they become major problems
Actions:
Pattern Categories:
| Pattern Type | Example Signal | Reflective Interpretation |
|---|---|---|
| Strategic Drift | README ≠ commit activity | Vision disconnection or premature pivoting |
| Documentation Inflation | Many similar .md files | Over-planning; avoidance through writing |
| Suppression Pattern | Many eslint-disable or any | Time pressure or fatigue |
| Frozen Dependencies | No updates for months | Fear of breakage or resistance to change |
| Commit Whiplash | Rapid reversals | External pressure or internal restlessness |
| Task-Reality Desync | Tests pass but features incomplete |
Validation:
Objective: Transform patterns into narrative insights with actionable interpretation
Actions:
Technical ↔ Human Translation Bridge:
| Technical Symptom | Human Parallel |
|---|---|
| Ignored build errors | Denial of fragility |
| Rapid branching | Loss of trust in direction |
| Over-commenting | Fear of being misunderstood |
| Skipped validation | Avoidance of feedback |
| Over-planning | Anxiety about imperfection |
| Unfinished refactors | Fatigue, low follow-through |
Key Principles:
Common Pitfalls:
Objective: Provide graduated recommendations based on urgency and impact
Actions:
Intervention Modes:
| Mode | Urgency | Action Type | Example |
|---|---|---|---|
| 🟢 OBSERVE | Low | Gentle awareness | "Velocity high, coherence low. Monitor for drift." |
| 🟡 REVIEW | Medium | Structured reflection | "Validate assumptions with peer. Check external feedback." |
| 🔴 HOLD | High | Pause + confirm | "Reality desynced. Reconcile before next build." |
Restoration Protocol (for HOLD items):
restoration:
1. pause: Create reflective checkpoint
2. align: Re-anchor to intent + artifact coherence
3. integrate: Capture insight in feedback loop
4. resume: Rebuild context from re-synced rhythm
Validation:
Objective: Calculate overall repository health score
Formula:
RCI = (Intent Alignment + Task Reality Sync + Technical Health) / 3
Scoring Components:
Intent Alignment (0-100):
Task Reality Sync (0-100):
Technical Health (0-100):
Interpretation:
| RCI Score | Status | Meaning |
|---|---|---|
| 85-100 | ✅ COHERENT | Rhythm aligned, healthy system |
| 70-84 | 🟡 MONITOR | Early drift present, watch closely |
| 50-69 | 🟠 MISALIGNED | Intent and reality diverging |
| <50 | 🔴 INCOHERENT | Rebuild context or reset rhythm |
Reveal, Not Judge Dark Matter Mode illuminates patterns without declaring them "right" or "wrong." Every pattern is an expression of system state. The goal is truthfulness, not perfection.
Interpretive, Not Diagnostic DMM doesn't predict outcomes — it provides reflective insights. Use uncertainty scores to maintain interpretive humility. The mirror knows when it is dreaming.
Coherence Over Speed Ask: "Does this make the system more truthful, not just faster?" Optimize for alignment between intent, artifact, and action.
Emotional Resonance Technical patterns reveal human patterns. Documentation inflation may indicate anxiety. Suppression patterns may show fatigue. Acknowledge the human element.
External Validation Internal consistency ≠ external utility. Tests may pass while users remain confused. Seek external feedback to validate interpretations.
When choosing analysis depth:
When choosing intervention mode:
If RCI < 50: HOLD (pause and reconcile)
If 50 ≤ RCI < 70: REVIEW (structured reflection)
If 70 ≤ RCI < 85: OBSERVE (gentle monitoring)
If RCI ≥ 85: MAINTAIN (celebrate and sustain)
When to recommend external validation:
If user discovered gap before internal analysis: HIGH PRIORITY
If documentation significantly exceeds implementation: RECOMMENDED
If team feels confused despite extensive docs: REQUIRED
Otherwise: NICE TO HAVE
# Dark Matter Mode Analysis
**Date:** [timestamp]
**Repository:** [name]
**RCI Score:** [score]/100 — [STATUS]
## Executive Summary
[3-5 sentence overview of key findings]
## Layer 1: Sensing — What Was Observed
[Code, documentation, temporal, environmental signals]
## Layer 2: Pattern Detection — The Weak Signals
[Identified patterns with evidence and intensity]
## Layer 3: Reflection — The Unseen, Unsaid, Unmeasured
[Narrative insights and interpretations]
## Layer 4: Action — Recommendations by Urgency
### 🔴 HOLD — Stop Before Proceeding
[Critical actions]
### 🟡 REVIEW — Reflection Requested
[Medium priority actions]
### 🟢 OBSERVE — Gentle Nudges
[Low priority awareness items]
## Repository Coherence Index
[Detailed scoring breakdown]
## Closing Reflection
[Supportive conclusion with path forward]
This skill integrates with dark-matter-analyzer-mcp for automated scanning:
// Run repository scan
await mcp.call('scan_repository', {
path: './my-project',
depth: 'deep',
include_patterns: ['*.md', '*.ts', '*.js'],
exclude_patterns: ['node_modules', 'dist']
})
// Calculate RCI
await mcp.call('calculate_rci', {
repository_path: './my-project'
})
// Generate report
await mcp.call('generate_report', {
format: 'markdown',
output_path: './dark_matter_report.md'
})
Signal: 30 documentation files averaging 543 lines each
Pattern: Documentation Inflation (HIGH)
Reflection:
"Planning activity significantly outpaces execution. The repository is drowning in its own self-awareness."
Action (HOLD):
Signal: 36 methodologies, 3 implementation tools (12:1 ratio)
Pattern: Execution Deficit (CRITICAL)
Reflection:
"The repository is a library of wisdom without hands to execute it. 92% of capabilities are aspirational, not actionable."
Action (HOLD):
Signal: All tests passing, but CLI has TODO comments, MCPs are placeholders
Pattern: Task-Reality Desynchronization (MODERATE)
Reflection:
"The repository passes its own tests but may not serve users. Validation ensures internal consistency, not external utility."
Action (REVIEW):
❌ Using Dark Matter Mode as a weapon - This is reflective, not punitive ❌ Over-relying on RCI score - It's orientation, not verdict ❌ Ignoring emotional subtext - Technical patterns reveal human state ❌ Skipping uncertainty scores - Maintain interpretive humility ❌ Making changes without user validation - Internal analysis ≠ external truth
"Dark Matter Mode remains a mirror — it does not predict, it illuminates."
Weekly Installs
0
Repository
GitHub Stars
18
First Seen
Jan 1, 1970
Security Audits
测试策略完整指南:单元/集成/E2E测试金字塔与自动化实践
11,200 周安装
| Validation of structure, not utility |
| Execution Deficit | Documentation >> Implementation | Hope-driven development |