Claude Skills Library

A collection of professional Claude Code skills for various domains and use cases.

📖 Documentation Site

https://takusaotome.github.io/claude-skills-library/

Browse the full skill catalog, getting-started guide, and per-skill usage guides on the GitHub Pages site.

Overview

This repository contains custom skills designed to extend Claude’s capabilities with specialized knowledge, workflows, and tools. Each skill is self-contained and follows best practices for skill development.

Repository Structure

claude-skills-library/
├── skills/                 # 119 published skills (with SKILL.md) — 119 dirs total
│   ├── data-scientist/
│   ├── project-manager/
│   ├── business-analyst/
│   └── ...
├── agents/                 # Agent definitions (18 agents)
│   ├── code-reviewer-bug-hunter.md
│   ├── document-reviewer-customer.md
│   └── ...
├── hooks/                  # Claude Code hook configurations
│   ├── current-datetime.json
│   └── notification-sound-macos.json
├── commands/               # Slash commands for Claude Code
│   └── clarify.md
├── skill-packages/         # Packaged .skill files for distribution
├── docs/                   # Additional documentation
└── README.md

Hooks

Useful hook configurations for Claude Code. See hooks/README.md for details.

Commands

Slash commands for Claude Code workflows.

/clarify - Plan Clarification

Resolves ambiguities in plan files through structured questioning using the AskUserQuestion tool.

Features:

• Detects vague language, missing specs, undefined constraints
• 2-4 questions per round with pros/cons options
• Records decisions with rationale and owner
• Updates plan file with decisions appendix
• Maximum 5 rounds (prevents infinite loops)

Installation: Copy commands/clarify.md to ~/.claude/commands/

Skill Catalog (119 Skills)

Note: every directory under skills/ is a published skill with a SKILL.md.

Business Strategy & Consulting (19 skills)

Project Management (8 skills)

Software Development & IT (21 skills)

Salesforce (4 skills)

Media Processing Tools (5 skills)

Documentation & Communication (24 skills)

QA & Testing (16 skills)

Compliance & Governance (12 skills)

Vendor Management (5 skills)

HR Management (2 skills)

Operations & Supply Chain (3 skills)

---

Agent Catalog (15 Agents)

Agents are specialized sub-agents that can be spawned by Claude Code using the Task tool. They run autonomously and return results.

Code Review Agents (4)

Document Review Agents (6)

Development & Analysis Agents (3)

Project & Vendor Management Agents (5)

---

Available Skills (Detailed)

✍️ Japanese Clear Writing

日本語の文書を、読み手が見つけ・理解し・使えるように書くためのスキルです。新規執筆、書き直し、添削、評価の4モードと、それらをまとめた統合モードを持ちます。

When to use:

• メモや箇条書きから、レポート・手順書・提案書・メールを書き起こすとき
• 既存の原稿やAI生成文を、意味を変えずに読みやすくしたいとき
• 原稿の問題点と修正案、またはルーブリックによる点数がほしいとき

Key Features:

• 依頼文と渡された原稿からモードを自動で選ぶ。指定があれば指定を優先する
• 意味・事実・数値・固有名詞・要件の強さを変えない。足りない情報は [要確認] で示す
• 指摘には必ず具体的な修正案を添える
• 6項目のルーブリック（目的・構成、文の明瞭さ、曖昧さ・正確さ、用語の一貫性、読み手適合性、表記・視認性）で採点し、scripts/score.py が合計点と判定を計算する
• references/standards.md に参照した規格・ガイドラインをまとめている

---

🛠️ STE Writer

英文を ASD-STE100（Simplified Technical English）の主要ルールに沿って書く、または書き換えるスキルです。手順書、マニュアル、SOP、ランブック、リリースノートが対象です。

When to use:

• 英語が母語でない読み手や機械翻訳を前提に、誤読のない英文の手順書を書きたいとき
• 既存の英文マニュアルを STE 寄りに書き換えたいとき

Key Features:

• 14の主要ルール（1文1指示、命令形、能動態、単純時制、1語1義、警告を先に置くなど）
• full（全ルール適用）と soft（「80% STE」程度）の2段階
• 公式の ASD-STE100 仕様書や辞書は同梱していない。出力は「STE準拠」ではなく「主要ルールに沿って記述」と表現する

---

📋 Task Runner

1ファイル1タスクのチケットを tasks/ 配下のフォルダで管理し、定期実行のたびに1件ずつ拾って実行し、結果をチケットに書き戻すスキルです。Cowork のスケジュール実行から1時間ごとなどで呼ぶ想定です。

When to use:

• 「タスクを1件処理して」「tasksフォルダをチェックして」など、溜めたタスクを順に進めたいとき
• タスク管理フォルダを初期設定したいとき、または看板ボードで状況を確認したいとき

Key Features:

• ステータスはフォルダ（todo/ doing/ blocked/ done/YYYY-MM/）だけで表し、ファイル内に status 欄を持たない
• doing/ をロック代わりに使い、タスクが残る間は次を始めない。60分超または開始日時が不明・未来時刻なら「停止の疑い」と表示し、終了確認後に人の指示で blocked/ へ移す
• dry-run / live の2モード。既定は dry-run
• scripts/build_dashboard.py が毎回 tasks/dashboard.html の看板ボードを再生成する
• assets/ にチケットのひな形、設定ファイル、サンプルタスク3件を同梱

---

🔍 Jev Artifact Style Review

「AIっぽい」という読後感を、修正できる文体・伝達品質の問題として指摘する日本語中心・英語対応のレビュースキルです。執筆者判定器ではありません。TypeSafe の Jev モデルが採点と根拠セグメントの選択を担当し、ホストのエージェントが原文を確認して具体的な修正文案を書きます。

When to use:

• 日本語・英語の文書、メール、提案書、技術資料、スライドが硬い・定型的・直訳調に見え、その理由を具体的に言語化したいとき
• 「もっと自然にして」ではなく「この一文をこう直す」という短いリストを作成担当者へ渡したいとき
• 改稿した版を同一条件で再評価し、本当に改善したかを確認したいとき

Key Features:

• 文体8軸と内容品質4軸を独立に採点し、ひとつの数値に混ぜない
• 0〜100の文体違和感指数は文体8軸のみを集計。AI執筆確率でも読者の違和感率でもない
• 用途別 profile 7種類（email / chat / report / proposal / technical / slides / formal）
• 外部送信前に --dry-run で送信内容を確認。dry run は採点せずダミースコアも出さない
• 引用文は Jev が選んだセグメントIDをもとに Python が原文から切り出す
• 改稿は原則2回まで。比較不能を「改善した」と解釈しない
• 対応形式は MD / TXT / HTML / JSON。追加依存で PDF / DOCX / PPTX のテキスト層も読む

Requirements: Python 3.10+、実際の採点には環境変数 TYPESAFE_API_KEY が必要です。

Status: 同梱テスト64件は合成応答によるもので、外部APIへ接続していません。実APIでの日本語精度・false positive・confidence の校正は未検証です。references/CALIBRATION.md の人手評価を実施するまで、指数は暫定値として扱ってください。

---

📊 Data Scientist

File: skill-packages/data-scientist.skill

A comprehensive data science workflow skill for analyzing tabular and time series data.

When to use:

• Conducting exploratory data analysis (EDA)
• Building predictive models (classification, regression)
• Performing feature engineering and selection
• Time series analysis and forecasting
• Generating data-driven insights and reports

Core Capabilities:

• ✅ Automated EDA with quality assessment and visualizations
• ✅ Feature engineering strategies (10+ technique categories)
• ✅ Model comparison across multiple algorithms
• ✅ Time series analysis with stationarity tests and forecasting
• ✅ Professional visualization templates
• ✅ Statistical rigor and industry best practices

Key Features:

Automated Scripts:

• auto_eda.py - Comprehensive automated exploratory data analysis
• model_comparison.py - Train and compare 7-9 ML algorithms
• timeseries_analysis.py - Specialized time series analysis and forecasting

Reference Guides:

• analysis_methodology.md - 7-phase data science workflow
• feature_engineering.md - 10 categories of feature engineering techniques
• evaluation_metrics.md - Complete guide to model evaluation metrics
• visualization_guide.md - Data visualization patterns and best practices

Templates:

• visualization_template.py - Professional chart and graph functions
• analysis_report_template.md - Comprehensive analysis report structure

Supported Problem Types:

• Classification (binary and multi-class)
• Regression
• Time series forecasting
• Clustering analysis

Example Use Cases:

• “Analyze this sales data and find key drivers”
• “Build a model to predict customer churn”
• “Forecast next quarter’s revenue based on historical data”
• “Compare different machine learning models for this classification problem”
• “Help me understand what features are most important”

Domain-Specific Features:

• Financial data analysis (technical indicators, ratios)
• E-commerce analytics (RFM, CLV, conversion metrics)
• Healthcare metrics (BMI, risk scores)

---

📋 Action Status Updater

File: skill-packages/action-status-updater.skill

A natural language action item tracking skill for managing status updates across communication channels.

When to use:

• Updating action item status via natural language (e.g., “Seanのメールには返信しておいた”, “Delegated to Mike”)
• Marking items as completed, delegated, deferred, or in-progress
• Tracking action items across email, Slack, meetings, and other channels
• Generating action item status reports
• Integrating with daily-comms-ops workflow

Core Capabilities:

• ✅ Natural language parsing for Japanese and English status updates
• ✅ Intent detection (completed, delegated, deferred, in-progress)
• ✅ Person and channel extraction from text
• ✅ Persistent YAML state with full history tracking
• ✅ Multiple export formats (daily-comms, slack, email)
• ✅ Fuzzy matching for description and person names

Key Features:

Automated Scripts:

• action_status_updater.py - CLI tool for action item management (init, add, update, report, export, list)
• nl_parser.py - Natural language parsing module for Japanese/English

Reference Guides:

• status_patterns.md - Comprehensive patterns for intent detection and target extraction
• integration_guide.md - Guide for daily-comms-ops workflow integration

Supported Update Patterns:

• Completed: 返信した, 完了, done, finished, sent
• Delegated: 〜に依頼, delegated to, assigned to
• Deferred: 延期, 来週, postponed, later
• In-Progress: 対応中, working on, in progress

Example Use Cases:

• “Seanのメールには返信しておいた” → Marks Sean’s email action as completed
• “Delegated the report to Mike” → Marks report as delegated to Mike
• “Lu対応予定” → Marks item as delegated to Lu

---

📋 Project Manager

File: skill-packages/project-manager.skill

A comprehensive PMBOK®-aligned project management skill for professional project delivery.

When to use:

• Defining project requirements (ISO/IEC/IEEE 29148 compliant)
• Reviewing project plans against PMBOK best practices
• Generating progress reports with Earned Value Management (EVM)
• Conducting comprehensive risk assessments
• Estimating project costs using industry-standard methods
• Assessing project health and recommending corrective actions

Core Capabilities:

• ✅ Requirements engineering with traceability matrix
• ✅ Earned Value Management (SPI, CPI, EAC, ETC, VAC, TCPI)
• ✅ Three-phase risk management framework (14 categories, 9 risk types)
• ✅ Cost estimation (analogous, parametric, bottom-up, three-point)
• ✅ Project plan reviews across 10 PMBOK knowledge areas
• ✅ Stakeholder management and communication planning

Key Features:

Automated Scripts:

• project_health_check.py - Automated project health assessment with scoring

Reference Guides:

• pmbok_knowledge_areas.md - Comprehensive guide to all 10 PMBOK knowledge areas
• risk_management_guide.md - Three-phase risk management methodology

Templates:

• requirements_definition_template.md - ISO/IEC/IEEE 29148 compliant requirements documentation
• progress_report_template.md - EVM-based progress reporting with comprehensive metrics
• risk_analysis_template.md - Structured risk analysis with probability/impact assessment

PMBOK Framework Coverage:

10 Knowledge Areas:

1. Integration Management
2. Scope Management
3. Schedule Management
4. Cost Management
5. Quality Management
6. Resource Management
7. Communications Management
8. Risk Management
9. Procurement Management
10. Stakeholder Management

5 Process Groups:

1. Initiating
2. Planning
3. Executing
4. Monitoring & Controlling
5. Closing

Example Use Cases:

• “Create a requirements definition document for our CRM implementation”
• “Review this project plan and identify gaps and risks”
• “Generate a progress report with EVM analysis - our SPI is 0.85”
• “Conduct a comprehensive risk assessment for this legacy migration project”
• “Estimate project costs using bottom-up estimation”
• “Analyze our project health based on current metrics”

Supported Methodologies:

• Traditional (Waterfall) project management
• Agile project management with PMBOK
• Hybrid approaches

---

🏢 AI-BPO Proposal Generator

File: skill-packages/ai-bpo-proposal-generator.skill

A comprehensive proposal generator for AI-powered BPO services tailored for Japanese companies operating in the US market.

When to use:

• Creating BPO service proposals for Japanese subsidiaries in the US (在米日系企業)
• Developing AI-enhanced outsourcing offerings for business processes
• Estimating ROI for AI implementation in back-office operations
• Building implementation roadmaps for phased AI adoption
• Generating bilingual (Japanese/English) proposal documents

Core Capabilities:

• ✅ Service module selection based on industry and pain points
• ✅ ROI calculation with NPV, payback period, and 3-year projections
• ✅ Phased implementation roadmap generation
• ✅ Bilingual proposal document generation (JA/EN)
• ✅ Industry-specific bundles with volume discounts

Key Features:

Automated Scripts:

• select_services.py - Service module selection based on industry and pain points
• calculate_roi.py - ROI calculation with current/future state cost analysis
• generate_roadmap.py - Implementation roadmap with milestones
• generate_proposal.py - Bilingual proposal document generator

Reference Guides:

• service-catalog.md - Complete AI-BPO service catalog with pricing (15 services, 3 bundles)
• client-intake-template.md - Bilingual client requirements questionnaire
• roi-methodology.md - ROI calculation methodology and benchmarks

Service Categories:

• Finance & Accounting (Invoice Processing, Expense Reports, Reconciliation, AR)
• HR & Payroll (Onboarding, Time Tracking, Payroll Processing)
• Customer Support (Ticket Routing, FAQ Automation)
• Data Processing (Document Digitization, Data Entry)
• Procurement (PO Processing, Vendor Management, Contract Analysis)

Example Use Cases:

• “Create a BPO proposal for ABC Manufacturing’s US subsidiary”
• “Calculate ROI for automating invoice processing (5000/month volume)”
• “Generate an implementation roadmap starting April 2025”
• “Build a bilingual proposal with full ROI analysis”

---

📊 Project Completeness Scorer

File: skill-packages/project-completeness-scorer.skill

A systematic project completeness evaluation skill that calculates weighted 0-100 scores across multiple dimensions.

When to use:

• Assessing project readiness for release or handoff
• Reviewing milestone deliverables against acceptance criteria
• Identifying gaps in code, documentation, or configuration
• Preparing for stakeholder reviews or gate approvals
• Scoring skill development projects within this repository

Core Capabilities:

• ✅ Multi-dimensional scoring (5 evaluation dimensions)
• ✅ Weighted scoring with customizable weights
• ✅ Gap identification with priority ranking
• ✅ Actionable next steps for each gap
• ✅ Project type templates (skill, webapp, library, document)

Key Features:

Automated Scripts:

• score_project.py - CLI tool for project evaluation with JSON/Markdown reports

Reference Guides:

• scoring-methodology.md - Detailed scoring rules and dimension definitions
• project-templates.md - Evaluation templates for different project types

Evaluation Dimensions:

• Functional Requirements (30%) - Core deliverables and features
• Quality Standards (20%) - Code quality, linting, formatting
• Test Coverage (25%) - Unit tests, integration tests
• Documentation (15%) - README, API docs, user guides
• Deployment Readiness (10%) - Config files, CI/CD, environment setup

Project Templates:

• skill - Claude Code skill projects
• webapp - Web applications (frontend/backend)
• library - Reusable libraries and packages
• document - Documentation-only projects
• custom - User-defined criteria from JSON file

Example Use Cases:

• “Score this skill project for completeness”
• “Evaluate our webapp before release”
• “Identify gaps in this library project”
• “Generate a completeness report for stakeholder review”

---

🔗 Project Artifact Linker

File: skill-packages/project-artifact-linker.skill

A comprehensive traceability and cross-referencing skill for project artifacts including WBS, meeting minutes, requirements, and decisions.

When to use:

• Tracing decisions back to the meetings where they were made
• Auditing project documentation for completeness and traceability
• Onboarding to a project and understanding decision history
• Preparing for project reviews with artifact linkage reports
• Extracting action items from meeting minutes and mapping to WBS tasks
• Generating compliance reports showing requirement-to-deliverable traceability
• Identifying orphaned artifacts (decisions without documentation, tasks without requirements)

Core Capabilities:

• Multi-document parsing (meetings, WBS, requirements, decisions)
• Automated link building with confidence scoring
• Bidirectional traceability matrix generation
• Gap detection for orphaned artifacts
• Overall health scoring for project traceability

Key Features:

Automated Scripts:

• parse_artifacts.py - Extract entities from project documents
• link_artifacts.py - Build cross-reference links between entities
• generate_traceability_report.py - Generate traceability matrix reports
• analyze_coverage.py - Identify gaps and orphaned artifacts

Reference Guides:

• artifact_patterns.md - Patterns for extracting entities from documents
• link_heuristics.md - Heuristics for establishing artifact links

Link Types:

• action_item → wbs_task - Action items mapped to implementing tasks
• decision → requirement - Decisions linked to requirements they address
• meeting → wbs_task - Meetings linked to tasks discussed
• requirement → wbs_task - Requirements traced to implementing tasks

Example Use Cases:

• “Parse all meeting minutes and link action items to WBS tasks”
• “Generate a traceability report for this project”
• “Find orphaned requirements without implementing tasks”
• “Show which decisions were made in which meetings”

---

🚀 Project Kickoff Bootstrapper

File: skill-packages/project-kickoff-bootstrapper.skill

新しいプロジェクトまたは既存リポジトリに Claude 用の kickoff 文脈（CLAUDE.md、docs/、.claude/rules/）を導入するスキル。リポジトリ証跡の自動検出とユーザー確認を組み合わせて、最小限の質問で最大限の文脈を整備する。

When to use:

• 新プロジェクト開始時に Claude の初期文脈を整備したい
• 既存リポジトリに後付けで CLAUDE.md と kickoff ドキュメント群を導入したい
• チーム間で完了判定やテスト方針の最低基準を揃えたい

Key Features:

• 8-phase bootstrap workflow (inspect → profile → question → resolve → create → extend → command → verify)
• 3 installation profiles (minimal, standard, full)
• 15 asset templates (CLAUDE.md, PROJECT_BRIEF, SKILL_ROUTING, QUALITY_GATES, TEST_STRATEGY, etc.)
• 6 reference guides (repository inspection, install profiles, question strategy, non-destructive update, consistency checklist, follow-on skill sequence)
• Non-destructive update policy for existing repositories
• Repository evidence inspection for auto-detection of stack, commands, and governance docs
• Follow-on skill recommendations (completion-quality-gate-designer, hidden-contract-investigator, etc.)

---

💼 Business Analyst

File: skill-packages/business-analyst.skill

A comprehensive BABOK®-aligned business analysis skill for professional requirements engineering and business process improvement.

When to use:

• Eliciting and documenting business requirements
• Conducting stakeholder analysis and engagement planning
• Analyzing and optimizing business processes
• Developing business cases with ROI/NPV analysis
• Performing gap analysis between current and future state
• Creating Business Requirements Documents (BRD)

Core Capabilities:

• ✅ Requirements elicitation with 6+ techniques (interviews, workshops, surveys, observation)
• ✅ Business process analysis with BPMN and value stream mapping
• ✅ Stakeholder analysis with Power/Interest matrix and RACI
• ✅ Business case development with financial analysis (ROI, NPV, IRR, Payback)
• ✅ Gap analysis with as-is/to-be process mapping
• ✅ Data quality assessment and profiling

Key Features:

Automated Scripts:

• business_analysis.py - Financial analysis, data profiling, and comparative analysis toolkit

Reference Guides:

• babok_framework.md - Complete BABOK® Guide v3 reference with 6 knowledge areas
• process_data_analysis.md - Process analysis and data assessment methodologies

Templates:

• business_requirements_document_template.md - ISO/IEC/IEEE 29148 compliant BRD
• business_case_template.md - Professional business case with financial analysis
• stakeholder_analysis_template.md - Comprehensive stakeholder management framework

BABOK Framework Coverage:

6 Knowledge Areas:

1. Business Analysis Planning & Monitoring
2. Elicitation & Collaboration
3. Requirements Life Cycle Management
4. Strategy Analysis
5. Requirements Analysis & Design Definition
6. Solution Evaluation

Example Use Cases:

• “Conduct stakeholder analysis for our digital transformation initiative”
• “Create a business case for automating our order processing system”
• “Document business requirements for the new customer portal”
• “Analyze our current sales process and identify optimization opportunities”
• “Perform gap analysis between current state and target operating model”
• “Profile this dataset to assess data quality for our analytics project”

Supported Methodologies:

• BABOK® Guide v3 standards
• Lean Six Sigma process improvement
• BPMN process modeling
• Value stream mapping
• MoSCoW prioritization

---

📊 Data Visualization Expert

File: skill-packages/data-visualization-expert.skill

A professional data visualization skill specialized in creating reader-friendly, accessible, and aesthetically pleasing charts and dashboards.

When to use:

• Creating any type of chart or graph
• Choosing the right visualization for your data
• Designing color schemes that work for everyone (including colorblind viewers)
• Creating executive dashboards or operational monitors
• Improving readability and aesthetic appeal of visualizations
• Preparing visualizations for presentations or reports
• Ensuring accessibility and WCAG compliance

Core Capabilities:

• ✅ Chart selection guidance (30+ chart types with decision trees)
• ✅ Colorblind-safe palettes (Okabe-Ito, Viridis, custom palettes)
• ✅ Professional dashboard design (Strategic, Operational, Analytical)
• ✅ Accessibility compliance (WCAG 2.1 standards)
• ✅ Story-driven visualization techniques
• ✅ Typography and layout principles
• ✅ Color theory and psychology

Key Features:

Automated Scripts:

• create_visualization.py - Command-line tool for creating professional charts (6 types: bar, line, scatter, heatmap, distribution, dashboard)

Reference Guides:

• visualization_principles.md - Design principles, color theory, typography, storytelling
• chart_selection_guide.md - Comprehensive guide to 30+ chart types with decision trees
• dashboard_design.md - Dashboard layout, KPI cards, interactivity patterns

Templates & Assets:

• visualization_templates.py - Ready-to-use Python templates (KPI cards, executive summaries, waterfall charts, etc.)
• color_palettes.json - 50+ professional color palettes (qualitative, sequential, diverging, business-specific, accessible)

Visualization Workflows:

1. Choosing the Right Chart Type
2. Applying Color Best Practices
3. Designing Professional Dashboards
4. Creating Story-Driven Visualizations
5. Ensuring Accessibility and Readability

Example Use Cases:

• “Create a bar chart comparing sales by region”
• “What’s the best way to visualize this time series data?”
• “Design a dashboard showing our key metrics”
• “Choose colors that work for colorblind people”
• “Make this chart more readable for a presentation”
• “Create visualizations that tell a compelling story”

Supported Chart Types:

• Comparison: Bar, Lollipop, Bullet
• Distribution: Histogram, Box Plot, Violin Plot
• Relationship: Scatter, Line, Heatmap
• Composition: Stacked Bar, Tree Map, Waterfall
• Time Series: Line, Area, Sparkline
• Geographic: Choropleth, Symbol Map
• Hierarchy: Tree Map, Sunburst

Accessibility Features:

• Colorblind-safe palettes (red-green colorblindness)
• WCAG 2.1 contrast ratios (4.5:1 text, 3:1 graphics)
• Alternative text patterns
• Responsive design for mobile/tablet/desktop

---

🔍 Vendor Estimate Reviewer

File: skill-packages/vendor-estimate-reviewer.skill

A comprehensive skill for evaluating vendor estimates for software development projects, from the client’s perspective. This skill helps you determine if a vendor’s cost estimate, timeline, and approach are reasonable and whether the project is likely to succeed.

When to use:

• You’ve received a vendor estimate/quotation for software development
• You need to compare multiple vendor estimates
• You want to validate if an estimate is reasonable before contract signing
• You need to prepare negotiation points with a vendor
• You want to identify potential project risks early
• You need documentation for stakeholder approval

Core Capabilities:

• ✅ Scope completeness analysis (identify gaps and missing items)
• ✅ Cost reasonableness validation (compare against market standards)
• ✅ Risk identification (14+ critical red flags, 60+ risk factors)
• ✅ Project feasibility assessment (timeline, team composition)
• ✅ Contract terms review (payment terms, warranties, IP ownership)
• ✅ Automated analysis script (Excel, CSV, PDF parsing)
• ✅ Comprehensive Markdown reports with recommendations

Key Features:

5 Integrated Workflows:

1. Initial Review and Triage (quick red flag check)
2. Detailed Analysis and Assessment (12 dimensions)
3. Vendor Clarification Preparation (structured questions)
4. Final Review and Recommendation (go/no-go decision)
5. Decision Support and Follow-Up (stakeholder presentation)

Industry Benchmarks:

• Labor rate standards by role and region (North America, Europe, Asia Pacific)
• Phase distribution percentages (Requirements 10-15%, Design 15-20%, Development 40-50%, Testing 15-25%)
• Project size benchmarks by type (web app, mobile, e-commerce, ERP, API, data migration)
• Team composition standards (senior 20-30%, mid 40-50%, junior 20-30%)
• Contingency recommendations (10-20% typical, 15-25% high-risk)

Automated Analysis Script:

python scripts/analyze_estimate.py vendor_estimate.xlsx \
  --vendor "Acme Corp" \
  --project "CRM System" \
  --budget 500000 \
  --output review_report.md

Critical Red Flags Detected:

• Testing <10% of effort (critical quality risk)
• No contingency buffer (cost overrun risk)
• All resources at 100% allocation (unrealistic)
• Vague task descriptions
• Missing critical phases
• Rates 30%+ below market (hidden costs or low quality)
• Large upfront payment >30% (financial risk)

• 60% junior resources (quality risk)

• No change management process

Risk Assessment:

• 10 risk categories (scope, estimation, resources, technical, process, PM, contract, organizational, domain-specific)
• 60+ specific risk factors with probability, impact, and mitigation strategies
• Risk scoring framework (high/medium/low prioritization)
• Contingency recommendations based on risk profile

Deliverables:

• Executive summary (1-2 pages) with go/no-go recommendation
• Comprehensive review report (15-25 pages Markdown)
• Interactive checklist (10-section scoring with approval workflow)
• Clarification request document (structured vendor questions)
• Negotiation strategy memo (cost reduction and value-add opportunities)

Supported Formats:

• Excel (.xlsx, .xls)
• CSV
• PDF
• Structured text descriptions

What You Get:

• 3 comprehensive reference documents (350+ pages equivalent):
  • Review checklist (12 sections, 200+ items)
  • Cost estimation standards (11 sections with data tables)
  • Risk factors guide (60+ risks with mitigations)
• 1 Python analysis script (automated parsing and reporting)
• 2 Markdown templates (comprehensive report + interactive checklist)

Use Cases:

• Pre-contract vendor estimate evaluation
• Multi-vendor comparison and selection
• RFP/RFQ response analysis
• Contract negotiation preparation
• Project risk assessment
• Stakeholder approval documentation
• Budget validation and planning

Best For:

• Project managers evaluating vendor proposals
• Technical leads reviewing architecture and effort estimates
• Procurement teams comparing multiple bids
• Finance teams validating budgets
• Business analysts assessing scope completeness
• CTO/CTOs making vendor selection decisions

---

📝 Vendor RFQ Creator（ベンダー見積依頼書作成）

File: skill-packages/vendor-rfq-creator.skill

顧客の曖昧な要望を構造化し、ベンダーに対する明確な見積もり依頼書（RFQ）を作成するスキル。システム開発プロジェクトのRFQ作成に特化し、日本語をデフォルトとして英語にも対応。

A skill that transforms vague client requirements into comprehensive RFQ (Request for Quotation) documents for software development projects. Specializes in creating clear, complete RFQs that enable vendors to provide accurate estimates.

When to use:

• 顧客から曖昧または不完全な要件を受け取ったとき
• ベンダーに送付する正式なRFQを作成する必要があるとき
• 正確な見積もりに必要なすべての情報を確実に含めたいとき
• 組織全体でRFQ作成を標準化したいとき

Core Capabilities:

• ✅ 要件引き出し（5W1Hフレームワークを使用した体系的な質問）
• ✅ 要件の明確化と構造化（曖昧な要望を明確な仕様に変換）
• ✅ RFQ文書作成（マークダウン形式で専門的な見積依頼書を生成）
• ✅ 品質レビュー（完全性、明確性、一貫性の検証）
• ✅ 日本語・英語対応（日本語がデフォルト、英語にも対応可能）

Key Features:

4つの統合ワークフロー:

1. Requirements Elicitation（要件引き出し）: 構造化された質問を通じて顧客ニーズを抽出
2. Requirements Structuring（要件構造化）: 曖昧な要件を明確な仕様に変換
3. RFQ Document Creation（RFQ文書作成）: 専門的で包括的なRFQ文書を生成
4. Quality Review（品質レビュー）: ベンダーへの送付前に完全性を検証

包括的なチェックリスト (rfq_checklist_ja.md - 150+項目):

• プロジェクト概要（基本情報、スコープ、成果物）
• 機能要件（ユーザー要件、システム機能、統合要件）
• 非機能要件（性能、可用性、セキュリティ、拡張性、運用、UX）
• 技術要件（技術スタック、開発環境、標準、ライセンス）
• プロジェクト管理要件（スケジュール、予算、体制、品質、リスク、変更管理）
• 契約・法務要件（契約形態、IP、守秘義務、保証）
• 見積依頼特有の要件（見積書形式、評価基準、提出手続き）
• 品質チェック（明確性、完全性、実現可能性、検証可能性）

RFQテンプレート (rfq_template_ja.md - 400+行): 完全な9セクション構成の見積依頼書テンプレート:

1. プロジェクト概要
2. 要件詳細（機能・非機能・技術）
3. プロジェクト管理要件
4. 契約・法務要件
5. 見積依頼内容（標準化された見積書フォーマット付き）
6. 評価・選定基準
7. 提出要項
8. 注意事項
9. 問い合わせ先

要件引き出し手法:

• 5W1Hフレームワーク: Who, What, Where, When, Why, How
• MoSCoW優先順位付け: Must have, Should have, Could have, Won’t have
• 構造化された質問: カテゴリー別、目的明確、選択肢付き
• 前提条件の文書化: 合理的な前提、根拠、影響を記載

見積書フォーマットの標準化:

• WBS（作業分解構造）テンプレート
• 必須フェーズ: 要件定義、設計、実装、テスト、デプロイメント、PM、品質保証
• 必須カラム: タスク名、詳細、役割、工数（人日）、単価、小計
• コンティンジェンシー（10-20%）
• 前提条件・制約条件セクション

品質チェック機能:

• 明確性チェック: 曖昧な表現の検出と修正提案
• 完全性チェック: 必須項目の網羅確認
• 一貫性チェック: 数値の整合性、用語の統一
• 実現可能性チェック: 技術的実現可能性、予算・スケジュールの現実性

プロジェクトタイプ別ガイダンス:

• Webアプリケーション: レスポンシブデザイン、SSL/TLS、3秒応答、100同時接続
• モバイルアプリ: iOS/Android対応、オフライン動作、プッシュ通知、2秒応答
• 基幹システム: 99.9%可用性、バックアップ、監査ログ、権限管理
• データ基盤: ETL、TB級処理、BI連携、バッチ・リアルタイム処理

ベストプラクティス:

1. ビジネス価値から始める（WHYを明確に）
2. 数値で具体的に表現（「多くのユーザー」→「初期1,000名、3年後10,000名」）
3. MoSCoWで優先順位付け
4. 文脈を提供（現状、課題、目指す姿）
5. 見積フォーマットを標準化
6. ベンダーの提案を促す
7. 現実的な期待値を設定
8. Q&Aプロセスを計画

よくある落とし穴を回避:

• ❌ 過去のRFQをカスタマイズせずにコピー
• ❌ 技術的に過度に規定
• ❌ スコープ境界が曖昧
• ❌ 非機能要件の欠落
• ❌ 非現実的な予算・スケジュール
• ❌ 評価基準の欠如

Use Cases:

• ベンダー選定プロセスの準備
• 複数ベンダーからの見積取得（相見積）
• RFP/RFQ作成の標準化
• プロジェクト予算承認のための文書化
• 要件の明確化とステークホルダー合意形成

Best For:

• プロジェクトマネージャー（ベンダー選定を担当）
• 調達・購買部門（システム開発の発注）
• ビジネスアナリスト（要件定義サポート）
• IT部門マネージャー（外部委託管理）
• プロダクトオーナー（開発ベンダーへの要件伝達）

Output Format:

• Markdown形式のRFQ文書（すぐにベンダーに送付可能）
• 構造化された見積書フォーマットテンプレート
• チェックリストベースの品質確認レポート

---

📋 Vendor Procurement Coordinator（ベンダー調達コーディネーター）

File: skill-packages/vendor-procurement-coordinator.skill

RFQ作成・送信から、ベンダー回答追跡、見積比較、クライアント向け見積作成までのエンドツーエンドのベンダー調達ワークフローを統合管理するスキル。vendor-rfq-creatorとvendor-estimate-creatorスキルをオーケストレーションし、メール自動化とステータス追跡を追加。

A skill that orchestrates the complete vendor procurement lifecycle from initial RFQ creation through vendor response tracking to final client-facing estimate generation. Integrates with existing vendor-rfq-creator and vendor-estimate-creator skills while adding email automation, response tracking, and procurement status management capabilities.

When to use:

• 単一プロジェクトで複数ベンダーへの見積依頼を管理するとき
• ベンダーの見積回答と期限を追跡するとき
• 受領したベンダー見積をクライアント向け見積に変換するとき
• RFQ配信をベンダーメールリストに自動化するとき
• エンドツーエンドの調達パイプラインを調整するとき

Core Capabilities:

• ✅ 調達プロジェクト初期化（ディレクトリ構造、設定ファイル作成）
• ✅ ベンダー管理（追加、編集、削除、CSVインポート）
• ✅ RFQメール送信（テンプレート、プレビュー/送信モード）
• ✅ 回答追跡（見積受領ログ、ステータス管理、リマインダー）
• ✅ 見積比較レポート（価格分析、納期比較、評価スコア）
• ✅ クライアント見積作成（マークアップ適用、統合）

Key Features:

8ステップ統合ワークフロー:

1. Initialize Procurement（調達プロジェクト初期化）
2. Create RFQ（RFQ文書作成 - vendor-rfq-creator連携）
3. Register Vendors（ベンダー登録）
4. Send RFQ（RFQ送信）
5. Track Responses（回答追跡）
6. Compare Quotes（見積比較）
7. Create Client Estimate（クライアント見積作成 - vendor-estimate-creator連携）
8. Generate Report（調達レポート生成）

調達ステータス管理:

• プロジェクトステータス: initialized → rfq_sent → quotes_received → evaluation → completed
• ベンダーステータス: pending → contacted → quote_received → selected/declined

メールテンプレート:

• RFQ送付メール（日本語/英語）
• リマインダーメール（1週間前/3日前）

Bundled Resources:

• references/procurement_workflow_guide.md: 調達プロセスガイド（6フェーズ）
• references/vendor_evaluation_criteria.md: ベンダー評価基準（6次元）
• assets/email_templates/: メールテンプレート集

Use Cases:

• 複数ベンダーからの相見積取得と管理
• 調達プロセスの標準化と監査証跡
• ベンダー選定の公平性と透明性確保
• クライアント向け見積作成の効率化

Best For:

• プロジェクトマネージャー（ベンダー選定担当）
• 調達・購買部門
• ITマネージャー
• システム開発ディレクター

Output Format:

• YAML形式の調達ステータス（procurement.yaml）
• Markdown形式のベンダー比較レポート
• Markdown形式のクライアント見積書

---

🎫 Vendor Support Ticket Tracker（ベンダーサポートチケット管理）

File: skills/vendor-support-ticket-tracker/

複数ベンダー（STX、Dell、HP等）のサポートチケットおよびRMAケースを一元管理するスキル。チケット状態、通信タイムライン、保留アクション、エスカレーション状況を追跡。ベンダーからのメール応答を解析してチケットステータスを自動更新し、フォローアップが必要なチケットを特定するステータスレポートを生成。

A skill for tracking vendor support tickets and RMA cases across multiple vendors (STX, Dell, HP, etc.). Maintains ticket state, communication timeline, pending actions, and escalation status. Integrates with email to auto-update ticket status from vendor responses and generates status reports identifying tickets requiring follow-up.

When to use:

• 新規ベンダーサポートチケットまたはRMAケースを登録するとき
• ベンダーからのメール返信を基にチケットステータスを更新するとき
• オープンチケットのステータスレポートを生成するとき
• フォローアップが必要な古いチケットを特定するとき
• 特定チケットのエスカレーション履歴を確認するとき
• ベンダー間のSLA遵守状況を追跡するとき

Core Capabilities:

• ✅ チケットデータベース初期化（YAML形式）
• ✅ チケット作成（ベンダー、ID、件名、優先度、カテゴリ）
• ✅ ステータス更新とタイムライン記録
• ✅ エスカレーション管理（理由、履歴追跡）
• ✅ 古いチケット検出（設定可能な閾値）
• ✅ ステータスレポート生成（Markdown形式）

Key Features:

8段階チケット状態マシン:

1. open - 新規作成、ベンダー確認待ち
2. acknowledged - ベンダーが受領確認
3. in-progress - ベンダーが対応中
4. awaiting-parts - RMA: 交換部品待ち
5. awaiting-customer - ベンダーが顧客対応待ち
6. escalated - 上位サポートにエスカレート
7. resolved - 問題解決、確認待ち
8. closed - 解決確認済み、完了

SLA管理:

• 優先度別SLA目標（critical: 4h, high: 24h, medium: 72h, low: 5日）
• SLA違反警告と自動エスカレーション
• 営業時間計算（9:00-18:00、休日除外）

Bundled Resources:

• references/ticket-lifecycle.md: チケット状態遷移、SLAガイドライン、エスカレーション手順
• scripts/ticket_manager.py: CLI（init/create/update/report/stale/escalate/show/list）

Use Cases:

• ハードウェア障害のRMAケース追跡
• ソフトウェアサポートチケット管理
• SLA遵守状況のモニタリング
• ベンダー対応のエスカレーション管理

Best For:

• ITサポートマネージャー
• システム管理者
• 調達・資産管理担当者
• ヘルプデスク運用者

Output Format:

• YAML形式のチケットデータベース（tickets.yaml）
• Markdown形式のステータスレポート

---

💰 Vendor Estimate Creator（ベンダー見積書作成）

File: skill-packages/vendor-estimate-creator.skill

RFQや要件を分析し、WBS作成、工数見積、コスト計算、ROI分析を行い、プロフェッショナルな見積書を生成するスキル。システム開発プロジェクトの見積作成に特化し、日本語をデフォルトとして英語にも対応。

A skill that transforms RFQs or project requirements into comprehensive cost estimates and quotations for software development projects. Specializes in creating accurate estimates with WBS, effort calculations, cost breakdowns, and ROI analysis to justify investments.

When to use:

• RFQや要件書を受け取り、見積書を作成する必要があるとき
• プロジェクトの工数とコストを高精度で算出したいとき
• 投資対効果（ROI）分析を提供し、投資を正当化したいとき
• 組織全体で見積作成を標準化したいとき
• クライアントのRFQに対してプロフェッショナルな見積書で応答するとき

Core Capabilities:

• ✅ RFQ分析と要件抽出（プロジェクト規模、複雑度、リスク評価）
• ✅ WBS作成とタスク特定（7フェーズ、タスク分解、完全性検証）
• ✅ 工数見積もり（4手法、標準工数適用、調整係数、検証）
• ✅ コスト計算と積み上げ（役割別単価、コスト集計、運用保守費算出）
• ✅ ROI分析とビジネスケース（As-Is/To-Be、ROI/NPV/IRR/回収期間、感度分析）
• ✅ 見積書生成（マークダウン形式、12セクション、品質チェック）

Key Features:

6つの統合ワークフロー:

1. RFQ Analysis and Understanding（RFQ分析と理解）: 要件抽出、規模測定、リスク評価
2. Work Breakdown and Task Identification（作業分解とタスク特定）: WBS作成、フェーズ定義、完全性検証
3. Effort Estimation（工数見積もり）: 手法選択、標準工数適用、調整係数、検証
4. Cost Calculation and Aggregation（コスト計算と積み上げ）: 役割別単価、コスト算出、集計
5. ROI Analysis and Business Case（ROI分析とビジネスケース）: 現状分析、期待効果、財務指標、感度分析
6. Estimate Document Generation（見積書生成）: テンプレート活用、全セクション記入、品質チェック

見積手法ガイド (estimation_methodology.md):

• 4つの見積手法（類推法、パラメトリック法、ボトムアップ法、三点見積もり）
• 見積精度とプロジェクトフェーズ
• プロジェクトタイプ別工数配分比率（Web、モバイル、基幹、API、データ基盤）
• 工数調整係数（複雑度、習熟度、技術リスク）
• コンティンジェンシー設定（5-40%、リスクレベル別）
• 人材リソース工数単価（役割別、経験年数別）
• 見積ベストプラクティスと注意事項

工数見積基準ガイド (effort_estimation_standards.md):

• 役割別生産性指標（LOC、FP）
• タスク別標準工数（要件定義、設計、実装、テスト、デプロイ、PM、QA）
• プロジェクトタイプ別標準工数（小規模、中規模、大規模）
• 調整係数の適用方法
• 工数見積の検証手法（生産性チェック、類似PJ比較、フェーズ比率チェック）

ROI分析ガイド (roi_analysis_guide.md):

• 主要財務指標（ROI、NPV、IRR、投資回収期間）の計算式と解釈
• ベネフィット分類（定量的：コスト削減、売上増加、生産性向上 / 定性的：顧客満足度、従業員満足度）
• ビジネスケース作成手順（As-Is分析、To-Be定義、投資額算出、キャッシュフロー分析、感度分析）
• ビジネスケース文書テンプレート
• ROI分析ベストプラクティス（保守的見積、定量化、複数シナリオ）

見積書テンプレート (estimate_template_ja.md - 400+行、12セクション): 完全な12セクション構成の見積書テンプレート:

1. エグゼクティブサマリー（プロジェクト概要、見積総額、ROI概要）
2. 前提条件（スコープ、技術前提、プロジェクト前提、納品物）
3. 見積詳細（フェーズ別見積、WBS詳細、役割別工数・単価）
4. プロジェクトスケジュール（マイルストーン、ガントチャート）
5. ROI分析（現状分析、期待効果、財務指標、感度分析）
6. チーム体制（体制図、主要メンバースキルセット）
7. リスクと対策（主要リスク、リスク軽減策）
8. 運用保守費用（保守内容、SLA）
9. 支払条件（開発費支払スケジュール、運用保守費支払）
10. 契約条件（知的財産権、守秘義務、瑕疵担保、変更管理）
11. その他（前提条件、除外事項、有効期限）
12. 承認（顧客・ベンダー署名欄）

見積手法:

• 類推法: 過去の類似プロジェクトベース（精度±25〜50%、構想段階）
• パラメトリック法: 画面数・API数等のパラメータベース（精度±15〜30%、企画段階）
• ボトムアップ法: WBSベースの積み上げ（精度±5〜15%、要件定義後）
• 三点見積もり: 楽観値・最頻値・悲観値による確率的見積（精度±10〜20%）

工数配分比率（Webアプリケーション標準）:

• 要件定義: 10-15%
• 設計: 15-20%
• 実装: 40-50%
• テスト: 20-25%
• デプロイ・運用準備: 5-10%
• プロジェクト管理: 10-15%
• 品質保証: 7-11%
• コンティンジェンシー: 10-25%

役割別標準単価（日本・2025年基準）:

• プロジェクトマネージャー: 100,000〜150,000円/人日
• アーキテクト: 90,000〜140,000円/人日
• シニアエンジニア: 80,000〜120,000円/人日
• ミドルエンジニア: 60,000〜90,000円/人日
• ジュニアエンジニア: 40,000〜65,000円/人日

ROI財務指標:

• ROI: (総利益 - 総投資額) / 総投資額 × 100
• NPV: 将来キャッシュフローの現在価値合計
• IRR: NPVをゼロにする割引率
• 投資回収期間: 初期投資を回収するまでの期間

ベストプラクティス:

1. 複数手法で見積を検証（ボトムアップ、パラメトリック、類似PJ比較）
2. 保守的に見積もり（便益は控えめ、コストは余裕を持って）
3. 前提条件を明記（すべての前提を文書化）
4. リスクを早期特定（技術、要件、統合、チーム、PM）
5. ROI分析を提供（ROI、NPV、IRR、回収期間、感度分析）
6. PM・QA工数を必ず計上（PM 10-15%、QA 7-11%）

よくある落とし穴を回避:

• ❌ コンティンジェンシーを含めない
• ❌ プロジェクト管理工数を忘れる
• ❌ データ移行を見落とす
• ❌ 非機能要件（性能、セキュリティ）を軽視
• ❌ 統合テスト工数を過小評価
• ❌ ROI分析で楽観的すぎる便益見積もり

Use Cases:

• RFQ応答のための見積書作成
• 社内プロジェクトの予算承認申請
• 複数提案の中から最適案選定（ROI比較）
• 投資判断のための財務分析
• ベンダー選定プロセスでの自社見積提出

Best For:

• ベンダー営業（RFQ応答、提案書作成）
• プロジェクトマネージャー（プロジェクト計画、予算申請）
• アカウントマネージャー（顧客提案、価格交渉）
• 経営企画部門（投資判断、ROI分析）
• 財務部門（予算承認、投資対効果評価）

Output Format:

• Markdown形式の見積書（12セクション、WBS詳細、ROI分析含む）
• 標準化されたWBS構造（タスクID、役割、工数、単価、小計）
• 財務指標付きROI分析（ROI、NPV、IRR、回収期間、感度分析）

---

📋 Project Plan Creator（プロジェクト計画作成）

File: skill-packages/project-plan-creator.skill

システム開発・導入プロジェクトの包括的なプロジェクト計画を作成する専門スキル。Project Charter、WBS、Ganttチャート、RACI matrix、リスク分析等をMarkdown + Mermaidで文書化し、PMBOK準拠のプロジェクト管理成果物を生成。

A specialist skill for creating comprehensive project plans for system development or implementation projects. Generates PMBOK-aligned project management artifacts including Project Charters, WBS, Gantt Charts, RACI matrices, and Risk analysis using Markdown + Mermaid visualizations.

When to use:

• 新規プロジェクトを開始し、プロジェクト計画を作成する必要があるとき
• Project Charterでプロジェクトを正式に承認・権限付与したいとき
• WBS、スケジュール、リソース配分を含む包括的なプロジェクト計画が必要なとき
• Mermaidダイアグラム（Gantt、WBS、ワークフロー）でプロジェクト構造を可視化したいとき
• PMBOK準拠のプロジェクト文書が必要なとき
• knowledge/pm-knowledgeフォルダのPM知識を活用したいとき

Core Capabilities:

• ✅ プロジェクトチャーター作成（12セクション、PMBOK準拠、正式承認文書）
• ✅ スコープ定義と管理（In/Out定義、WBS、スコープ変更管理プロセス）
• ✅ スケジュール作成（Mermaid Ganttチャート、依存関係、クリティカルパス）
• ✅ リソース計画とRACI Matrix（体制図、役割定義、責任明確化）
• ✅ リスク管理計画（リスク識別、分析、対応計画、監視プロセス）
• ✅ コミュニケーション・品質計画（コミュニケーションマトリックス、品質基準、QAプロセス）
• ✅ 統合と文書生成（すべての計画成果物を包括的プロジェクト計画に統合）

Key Features:

7つの統合ワークフロー:

1. Project Charter Creation（プロジェクトチャーター作成）: インプット収集、目的定義、スコープ、成果物、予算、ステークホルダー、リスク、成功基準
2. Scope Definition and Management（スコープ定義と管理）: スコープステートメント、スコープ境界可視化、WBS作成、スコープベースライン、変更管理プロセス
3. Schedule Development with Gantt Charts（Ganttチャートでスケジュール作成）: アクティビティ定義、シーケンス、期間見積、Mermaid Gantt作成、クリティカルパス特定、スケジュール最適化
4. Resource Planning and RACI Matrix（リソース計画とRACI Matrix）: プロジェクト役割定義、RACI Matrix作成、コミュニケーションプロトコル、チーム構造可視化
5. Risk Management Planning（リスク管理計画）: リスク識別（5カテゴリ）、定性的分析、対応計画、監視プロセス確立
6. Communication and Quality Planning（コミュニケーション・品質計画）: ステークホルダー分析、コミュニケーションマトリックス、品質基準、QAプロセス、受け入れ基準
7. Integration and Document Generation（統合と文書生成）: テンプレート活用、全セクション記入、Mermaidダイアグラム生成、品質チェック、承認取得

プロジェクトチャーターガイド (project_charter_guide.md):

• PMBOK準拠のチャーター作成ガイド（12セクション）
• プロジェクト基本情報、エグゼクティブサマリー、背景と目的
• スコープ（In/Out）、主要成果物、マイルストーン
• 概算予算（開発、インフラ、間接費）
• 前提条件と制約、主要リスク
• ステークホルダー、成功基準、承認
• システム開発プロジェクトの実例
• ベストプラクティスとよくある落とし穴

プロジェクト計画テンプレート (project_plan_template.md - 400+行、12セクション): Mermaidダイアグラム5つを含む完全なプロジェクト計画テンプレート:

1. プロジェクト基本情報（名称、コード、PM、期間、予算）
2. エグゼクティブサマリー（概要、目的、スコープ、成果物、期間、予算）
3. プロジェクトスコープ（スコープ記述、スコープ境界Mermaid図）
4. WBS（WBS階層Mermaid図、詳細WBS表）
5. プロジェクトスケジュール（Mermaid Ganttチャート、マイルストーン、クリティカルパス）
6. 成果物一覧（成果物、説明、納期、承認者、受け入れ基準）
7. チーム体制とRACI（体制図Mermaid、RACI Matrix）
8. リスク管理計画（リスク登録簿、対応計画、監視プロセスMermaid図）
9. コミュニケーション計画（コミュニケーションマトリックス、会議体、報告）
10. 品質管理計画（品質基準、QAプロセスMermaid図、レビュー体制）
11. 変更管理計画（変更管理プロセスMermaid図、CCB、影響評価）
12. 承認（PM、PO、Sponsor署名欄）

Mermaid可視化（5つのダイアグラム）:

• Scope Boundary Diagram: In Scope vs Out of Scope可視化（graph TB）
• WBS Hierarchy: 作業分解構成の階層図（graph TD）
• Gantt Chart: プロジェクトスケジュールとマイルストーン（gantt）
• Risk Monitoring Process: リスク監視プロセスフロー（graph LR）
• Change Management Process: 変更管理プロセスフロー（graph TD）

PMBOK知識エリア統合:

• Project Integration Management（統合管理）
• Project Scope Management（スコープ管理）
• Project Schedule Management（スケジュール管理）
• Project Cost Management（コスト管理）
• Project Quality Management（品質管理）
• Project Resource Management（リソース管理）
• Project Communications Management（コミュニケーション管理）
• Project Risk Management（リスク管理）
• Project Procurement Management（調達管理）
• Project Stakeholder Management（ステークホルダー管理）

リスク管理フレームワーク:

• 技術リスク: 新技術習得、性能要件、外部API変更
• 要件リスク: 要件変更頻発、ステークホルダー間認識相違、スコープクリープ
• リソースリスク: キーメンバー離脱、スキル不足、リソース競合
• 統合リスク: 既存システム連携困難、データ移行複雑、第三者システム互換性
• 外部リスク: ベンダー納期遅延、法規制変更、市場環境変化

RACI Matrix定義:

• R (Responsible): 実行責任者（作業を実施する人）
• A (Accountable): 説明責任者（最終意思決定者、各タスク1名のみ）
• C (Consulted): 相談先（事前に意見を聞く人）
• I (Informed): 報告先（結果を報告する人）

スケジュール見積手法:

• PERT三点見積もり: (楽観値 + 4×最頻値 + 悲観値) / 6
• 依存関係: FS（Finish-to-Start）、SS（Start-to-Start）、FF（Finish-to-Finish）、SF（Start-to-Finish）
• クリティカルパス: 最長パスの特定と管理集中
• スケジュール短縮: Crashing（リソース追加）、Fast Tracking（並行実施）

品質基準例:

• コード品質: カバレッジ80%以上、複雑度10以下、コードレビュー、静的解析
• テスト品質: 単体テスト80%、統合テスト全API、システムテスト全機能、UAT全ビジネスシナリオ
• ドキュメント品質: 全設計書レビュー・承認、API仕様自動生成、運用マニュアルユーザーレビュー

ベストプラクティス:

1. チャーターから開始（プロジェクト正式承認、PM権限付与）
2. ビジュアル図を活用（MermaidでWBS、Gantt、プロセスフロー可視化）
3. RACIを早期定義（役割と責任の明確化、混乱回避）
4. プロアクティブなリスク管理（早期識別、継続監視）
5. すべてベースライン化（スコープ、スケジュール、コストのベースライン確立）
6. バージョン管理（文書バージョンと変更履歴維持）
7. ステークホルダーエンゲージメント（定期的コミュニケーション、期待値管理）
8. PMBOK準拠（一貫性とプロフェッショナリズム）

よくある落とし穴を回避:

• ❌ プロジェクトチャーター作成をスキップ
• ❌ 曖昧なスコープ定義（スコープクリープ発生）
• ❌ バッファなしの非現実的スケジュール
• ❌ RACI Matrixなし（役割混乱）
• ❌ 問題発生まで リスク管理を無視
• ❌ 変更管理プロセスの欠如
• ❌ ステークホルダーコミュニケーション不足
• ❌ 品質基準を事前に設定しない

Use Cases:

• 新規システム開発プロジェクト立ち上げ
• システム導入プロジェクトの計画策定
• 大規模リプレースプロジェクトの計画
• マルチベンダープロジェクトの統合計画
• PMO標準プロジェクト計画テンプレート作成
• プロジェクト提案書のプロジェクト計画セクション

Best For:

• プロジェクトマネージャー（プロジェクト計画作成、承認取得）
• PMO（標準テンプレート、プロセス確立）
• プロダクトオーナー（スコープ定義、優先順位）
• エンタープライズアーキテクト（技術計画、統合設計）
• ステアリングコミッティ（計画レビュー、承認）

Output Format:

• Markdown + Mermaid形式のプロジェクト計画（12セクション）
• 5つのMermaidダイアグラム（Scope、WBS、Gantt、Risk Process、Change Management）
• RACI Matrix（全タスク・成果物の責任明確化）
• リスク登録簿（リスク識別、分析、対応策）
• コミュニケーション・品質計画（マトリックス、基準、プロセス）

Framework Alignment:

• PMBOK® Guide 6th Edition準拠
• PMBOK® Guide 7th Edition principles対応
• ISO 21500プロジェクトマネジメント標準
• Prince2® compatible processes

---

📊 WBS Review Assistant（WBS照合レビュー支援）

File: skill-packages/wbs-review-assistant.skill

Work Breakdown Structure (WBS) Excelファイルをプロジェクト要件書、ヒアリングシート、過去の決定事項と照合し、ギャップ、不整合、決定事項からの逸脱を自動検出。Excelセルに直接レビューコメントを追加し、優先順位付けされたサマリーレポートを生成する専門スキル。

A specialist skill for reviewing and annotating Work Breakdown Structure (WBS) Excel files against project requirements, hearing sheets, and prior decisions. Automatically identifies gaps, inconsistencies, and deviations. Adds review comments directly to Excel cells and generates prioritized summary reports.

When to use:

• WBSを要件書と照合してレビューする必要があるとき
• WBS + 要件ドキュメントでギャップ分析が必要なとき
• ヒアリングシートの決定事項とWBSの整合性を確認したいとき
• ベースライン承認前にWBSの完全性を検証したいとき
• Excelにレビューコメントをアノテーションしたいとき
• キックオフ前にWBS課題の優先順位付けリストが必要なとき

Core Capabilities:

• ✅ 要件トレーサビリティ分析（要件IDとWBSタスクのマッピング）
• ✅ 構造検証（WBSコード階層、フェーズ構成、マイルストーン）
• ✅ コンテンツ品質チェック（工数見積、タスク記述、受入基準）
• ✅ ヒアリングシート決定事項クロスチェック（決定がWBSに反映されているか）
• ✅ Excel直接アノテーション（セルコメント + 条件付き書式で重要度可視化）
• ✅ 優先順位付けレポート生成（JSON、Markdown、アノテーション済Excel）

Key Features:

5段階レビュープロセス:

1. Document Ingestion（ドキュメント取込）: WBS Excel、要件書、ヒアリングノート解析
2. Traceability Analysis（トレーサビリティ分析）: 要件→WBSタスク、WBS→要件の双方向マッピング
3. Structural Validation（構造検証）: 階層整合性、WBSコード番号、フェーズ構成
4. Content Quality Checks（コンテンツ品質チェック）: 工数見積、タスク記述、品質ゲート
5. Finding Prioritization（所見優先順位付け）: 重要度 × 影響度で優先順位算出

チェックリストカテゴリ（YAML設定）:

• Completeness（完全性）: 全要件マッピング、成果物識別、受入基準、依存関係
• Consistency（一貫性）: WBSコード番号、工数見積、リソース割当、詳細レベル
• Alignment（整合性）: ヒアリングシート決定反映、スコープ境界遵守、技術選択整合
• Quality（品質）: テストフェーズ存在、ドキュメント作成タスク、レビューゲート
• Estimation（見積品質）: 積上計算正確性、妥当な工数範囲、バッファ/コンティンジェンシー

重要度レベル定義:

• Critical（致命的）: WBSベースライン不可（要件欠落、計算エラー）
• Major（重大）: プロジェクト成功に高リスク（テスト不足、詳細不足）
• Minor（軽微）: 改善機会（フォーマット、命名規則）

3つの出力形式:

1. アノテーション済Excel (wbs_annotated_YYYYMMDD_HHMMSS.xlsx):

   • 元のWBS構造保持（非破壊レビュー）
   • セルコメントに所見追加（Finding ID、重要度、推奨対応）
   • 条件付き書式（赤=Critical、オレンジ=Major、黄=Minor）
   • 新シート「Review Summary」（課題ダッシュボード、フィルタ機能）

2. Markdownサマリーレポート (wbs_review_summary_YYYYMMDD_HHMMSS.md):

   • エグゼクティブサマリー（重要度別課題数、準備度スコア）
   • 優先順位付け所見リスト（Top 10、推奨対応付き）
   • 要件カバレッジ分析（トレーサビリティマトリックス）
   • 欠落タスク候補（未マッピング要件から推奨）

3. JSONギャップ分析 (wbs_gaps_YYYYMMDD_HHMMSS.json):

   • 機械可読な所見データ（CI/CD統合可能）
   • トレーサビリティマトリックス（要件→WBSタスク）
   • 各チェック基準の検証結果

Readiness Score計算式:

Base Score = 100
- Criticalごとに20点減点
- Majorごとに5点減点
- Minorごとに1点減点
Readiness Score = max(0, Base Score - 減点合計)

90-100: ベースライン承認可
70-89: 要修正だが使用可
50-69: 重大なギャップ、大幅な作り直し必要
<50: 不完全、レビュー不可

よくあるWBS課題パターン（自動検出）:

• ❌ MISSING-REQ: 要件書に記載があるがWBSタスクなし（Critical）
• ❌ MISSING-MILESTONE: フェーズ境界に承認マイルストーンなし（Major）
• ❌ MISSING-TESTING: 単体テスト、結合テスト、UAT不足（Critical）
• ❌ INCONSISTENT-NUMBERING: WBSコード階層スキップ（1.1→1.3）（Major）
• ❌ MISSING-EFFORT: リーフタスクに工数見積なし（Major）
• ❌ ROLLUP-ERROR: 親タスク合計 ≠ 子タスク合計（Critical）
• ❌ HEARING-DECISION-MISSING: ヒアリングシート決定事項がWBSに未反映（Critical）
• ❌ OUT-OF-SCOPE-TASK: 要件書に記載ない作業がWBSに存在（Major）
• ❌ TECH-CHOICE-MISMATCH: 要件書と異なる技術スタック（Major）
• ❌ NO-ACCEPTANCE-CRITERIA: マイルストーンに受入基準なし（Major）
• ❌ MISSING-DOCUMENTATION: 成果物にドキュメント作成タスクなし（Major）
• ❌ NO-REVIEW-GATE: 開発タスクに正式レビューなし（Major）
• ❌ UNREALISTIC-ESTIMATE: タスク工数が大きすぎる（>80h、分解が必要）（Major）
• ❌ NO-CONTINGENCY: コンティンジェンシーバッファなし（Minor）

バイリンガルサポート:

• 日本語WBS対応（タスク名、要件ID: 要件-001、機能-002）
• 日本語ヒアリングノート対応（決定、合意、承認マーカー）
• バイリンガルレポート生成（EN/JA並列表示オプション）

Use Cases:

• システム開発プロジェクトのWBS品質保証
• ベースライン承認前のWBS最終チェック
• ベンダー提出WBSの妥当性検証
• 要件変更後のWBS更新確認
• プロジェクトキックオフ前のWBS準備度評価
• PMO標準レビュープロセスの自動化

Best For:

• プロジェクトマネージャー（WBS品質保証、ベースライン前確認）
• PMO（標準レビュープロセス、品質ゲート）
• ビジネスアナリスト（要件トレーサビリティ確認）
• QAリード（テスト計画妥当性、受入基準確認）
• ステアリングコミッティ（WBS承認前のリスク評価）

Framework Alignment:

• PMBOK® Guide (WBS作成、スコープ管理、品質保証)
• ISO 21500 (Work Breakdown Structure standards)
• IEEE 830 (Requirements Traceability)

---

🐛 Bug Ticket Creator（不具合チケット作成）

File: skill-packages/bug-ticket-creator.skill

システムテスト中に発見した不具合を包括的なバグチケットに変換する対話型スキル。ユーザーと対話しながら再現手順、環境情報、重要度判定など必要な情報を収集し、プロフェッショナルなマークダウン形式の不具合報告書を生成。

An interactive skill that transforms bug discoveries during system testing into comprehensive bug tickets. Guides users through systematic questioning to collect reproduction steps, environment details, severity assessment, and generates professional Markdown bug reports.

When to use:

• テスト中に不具合を発見し、バグチケットを作成する必要があるとき
• 再現手順を体系的に整理したいとき
• 適切な重要度・優先度を判定したいとき
• 標準化されたプロフェッショナルなバグ報告フォーマットが必要なとき
• 開発チームが迅速に理解・修正できる高品質なバグチケットを作成したいとき

Core Capabilities:

• ✅ 対話型の情報収集（6つのワークフロー、体系的な質問）
• ✅ 初期発見（何が、どこで、いつ発生したか）
• ✅ 再現手順収集（CLEAR原則準拠、事前条件、ステップバイステップ、再現率）
• ✅ 期待値vs実際の動作（仕様との差異明確化）
• ✅ 環境情報収集（OS、ブラウザ、デバイス、解像度、ネットワーク等）
• ✅ 重要度・優先度判定（Severity: Critical/High/Medium/Low、Priority: P0/P1/P2/P3）
• ✅ バグチケット生成（マークダウンファイル出力、日本語/英語対応）

Key Features:

6つの統合ワークフロー:

1. Initial Bug Discovery（初期発見）: 何が起きたか、どこで起きたか、いつ起きたか、バグタイプ分類、初期重要度評価
2. Reproduction Steps Collection（再現手順収集）: 事前条件確立、ステップバイステップ収集、失敗ポイント特定、再現率確認
3. Expected vs Actual Behavior（期待vs実際）: 期待される動作明確化、実際の動作記録、ギャップ分析、証拠収集（スクリーンショット、ログ）
4. Environment Information Collection（環境情報収集）: 基本環境情報（OS、ブラウザ、デバイス）、追加詳細（解像度、ネットワーク、拡張機能）、環境依存性確認
5. Severity and Priority Assessment（重要度・優先度判定）: Severity判定（技術的影響）、Priority判定（ビジネス緊急性）、組み合わせ説明
6. Bug Ticket Generation（バグチケット生成）: 残情報収集、テンプレート選択、情報入力、レビュー、マークダウンファイル生成

不具合分類ガイド (defect_classification_guide.md):

• 7つの不具合タイプ分類:
  • 機能不具合（Functional Defect）: 機能欠如、機能誤動作、余分な機能
  • UI/UX不具合（UI/UX Defect）: 表示崩れ、デザイン不一致、操作性問題、アクセシビリティ問題
  • パフォーマンス不具合（Performance Defect）: 応答速度問題、リソース消費問題、スケーラビリティ問題
  • データ不具合（Data Defect）: データ損失、データ不整合、データ破損
  • セキュリティ不具合（Security Defect）: 認証・認可問題、データ漏洩、インジェクション、セッション管理問題
  • 統合・連携不具合（Integration Defect）: API連携エラー、データ連携エラー、認証連携エラー
  • 環境依存不具合（Environment-Specific Defect）: ブラウザ依存、OS依存、デバイス依存、環境設定依存
• 発生フェーズ分類（要件定義、設計、実装、テスト、デプロイ）
• 原因分類（仕様理解不足、設計ミス、コーディングエラー、環境問題、データ問題、第三者システム問題）
• 影響範囲分類（全体影響、機能影響、局所影響、視覚的影響）

重要度・優先度判定ガイド (severity_priority_guide.md):

• 重要度（Severity） - 技術的影響（QAチームが判定）:
  • Critical: システム使用不可、データ損失、セキュリティリスク
  • High: 主要機能使用不可、回避策なし
  • Medium: 機能に問題あるが回避策あり
  • Low: 視覚的問題のみ、影響軽微
• 優先度（Priority） - ビジネス緊急性（PO/PMが判定）:
  • P0（最優先）: 本番環境で発生中、即座対応（数時間以内）
  • P1（高優先）: 1-3営業日以内の対応が必要
  • P2（中優先）: 次回リリースで対応（1-2週間）
  • P3（低優先）: バックログ追加、時間がある時に対応
• 重要度×優先度マトリックス（判定ガイド）
• 判定フローチャート
• 実際の判定例5例（ログイン不可、管理画面レポート不具合、スマホ表示崩れ、ヘルプ誤字、管理者機能バグ）
• 特殊ケース判定（セキュリティ、パフォーマンス、UI/UX、環境依存、顧客影響）

再現手順の書き方ガイド (reproduction_steps_guide.md):

• CLEAR原則:
  • Complete（完全）: すべての必要な情報
  • Logical（論理的）: 論理的な順序
  • Explicit（明示的）: 曖昧な表現なし
  • Actionable（実行可能）: 誰でも同じ手順を実行可能
  • Reproducible（再現可能）: 何度でも同じ結果
• 再現手順の基本構成（事前条件、再現手順、期待結果、実際結果、環境情報）
• 事前条件の書き方（ログイン状態、データ状態、システム状態）
• 再現手順の書き方（番号付け、一つの手順に一つの操作、具体的記述、操作対象明示、入力値記載）
• 期待結果・実際結果の書き方（具体的、観測可能、仕様書参照）
• 環境情報の書き方（OS、ブラウザ、デバイス、解像度、ネットワーク、言語設定、タイムゾーン）
• スクリーンショット・動画・ログの添付ガイド
• 良い例と悪い例の比較（実例2例: ログイン失敗、スマホ表示崩れ）

バグチケットテンプレート (日本語・英語): 完全な12セクション構成:

1. ヘッダー（タイトル、作成日、報告者、ステータス、Severity、Priority）
2. 不具合概要（Summary）
3. 不具合分類（Classification）: タイプ、サブカテゴリ、影響範囲、発生フェーズ、影響ユーザー数
4. 事前条件（Preconditions）
5. 再現手順（Steps to Reproduce）: ステップバイステップ
6. 期待される結果（Expected Result）
7. 実際の結果（Actual Result）
8. 再現頻度（Reproduction Rate）
9. 環境情報（Environment）: OS、ブラウザ、デバイス、解像度等
10. 添付ファイル（Attachments）: スクリーンショット、動画、ログ
11. 追加情報（Additional Information）: エラーメッセージ全文、コンソールログ、ネットワークログ、サーバーログ
12. 関連情報（Related Information）: 関連チケット、仕様書、デザインモック、PR
13. 暫定回避策（Workaround）
14. 備考・コメント（Notes/Comments）
15. 影響分析（Impact Analysis）: ビジネス影響、セキュリティ影響、ユーザー影響
16. 対応方針推奨（Recommended Action）: 推奨優先度、推奨対応方法
17. 更新履歴（Update History）
18. チェックリスト（Quality Check）: 報告者確認、開発チーム確認

対話の流れ（インタラクティブ質問）:

1. 初期発見フェーズ:
   - 何が起きましたか？
   - どこで起きましたか？
   - 問題のタイプは？（7つの選択肢）
   - 影響の大きさは？

2. 再現手順フェーズ:
   - 事前条件は？（ログイン状態、データ、開始画面）
   - 操作を一つずつ教えてください
   - どの時点で問題が発生しましたか？
   - 毎回発生しますか？（再現率）

3. 期待vs実際フェーズ:
   - 本来どうあるべきですか？
   - 実際に何が起きましたか？
   - スクリーンショットはありますか？

4. 環境情報フェーズ:
   - OS、ブラウザ、デバイスは？
   - 追加の環境詳細は？
   - 他の環境でも発生しますか？

5. 重要度・優先度フェーズ:
   - システム全体が使えませんか？
   - データ損失はありますか？
   - 本番環境で発生していますか？
   - リリース予定日は近いですか？

6. チケット生成フェーズ:
   - チケットタイトルを考えてください
   - 回避策はありますか？
   - 添付ファイルは？
   - 日本語？英語？

ベストプラクティス:

1. 一度に一つの質問（ユーザーを圧倒しない）
2. 明確化と確認（曖昧な回答にフォローアップ質問）
3. 技術知識を仮定しない（用語説明、例示）
4. 体系的な収集（ワークフロー順守、ステップスキップしない）
5. プロフェッショナルな出力（整形されたMarkdown、テンプレート厳守）
6. 証拠を奨励（スクリーンショット、動画、ログ依頼）

よくある落とし穴を回避:

• ❌ 曖昧な説明を受け入れる（「エラーが出る」で終わらせない）
• ❌ 再現頻度をスキップ（毎回？何回中何回？を確認）
• ❌ 不完全な環境情報（「Chrome」だけでなくバージョンまで）
• ❌ 複数操作を一つにまとめる（ステップを分解）
• ❌ 証拠を求めない（スクリーンショット・ログを依頼）

Use Cases:

• システムテスト中の不具合発見
• 結合テスト、システムテスト、受け入れテストでのバグ報告
• 本番環境でのインシデント報告
• QAチームによる不具合文書化
• テスト担当者のバグ報告スキル向上
• チーム内でのバグ報告標準化

Best For:

• QAエンジニア（不具合発見・報告）
• テスター（テスト実行・バグチケット作成）
• 開発者（自己テスト時のバグ記録）
• プロジェクトマネージャー（バグ管理品質向上）
• ビジネスアナリスト（UAT時のバグ報告）

Output Format:

• Markdown形式の不具合チケット（.md file）
• ファイル名形式: BUG-[NUMBER][short-description][YYYY-MM-DD].md
• 完全な12セクション構成
• プロフェッショナルなフォーマット
• JIRAやRedmine、GitHub Issues等のチケット管理システムにそのまま利用可能

Framework Alignment:

• ISTQB（International Software Testing Qualifications Board）準拠のバグ報告
• IEEE 829テスト文書化標準
• ISO/IEC/IEEE 29119ソフトウェアテスト標準

Bilingual Support:

• 日本語（デフォルト）: 対話、テンプレート、ガイド全て日本語対応
• English: Full English template and workflow support

---

📧 Internal Email Composer

File: skills/internal-email-composer/SKILL.md

Generate professional internal email drafts for common coordination tasks in business environments. Creates culturally-appropriate bilingual (Japanese/English) emails with proper business tone.

When to use:

• Drafting internal emails to request vendor quote compilation
• Forwarding RFQ documents to internal stakeholders
• Delegating tasks to team members with clear instructions
• Sending status update emails for ongoing projects
• Composing follow-up emails for pending responses
• Creating escalation emails for delayed deliverables

Core Capabilities:

• ✅ 6 email scenarios (Vendor RFQ, Task Delegation, Status Update, Follow-up, Escalation, Info Request)
• ✅ Bilingual support (Japanese/English) with culturally-adapted content
• ✅ 3 urgency levels with appropriate subject prefixes
• ✅ Template engine with variable substitution
• ✅ Business etiquette compliance (敬語, keigo for Japanese)

Key Components:

• scripts/compose_email.py - Main email composition script with CLI interface
• references/email-templates.md - Template patterns for each email type
• references/business-etiquette-guide.md - Cultural considerations for JA/EN emails

Output Formats:

• Markdown email draft with subject, greeting, body, closing, signature
• JSON structure for programmatic integration

Example Use Cases:

• “Create a Japanese email to forward RFQ to procurement team”
• “Draft a task delegation email for Q4 budget review”
• “Compose a follow-up email for pending vendor quotes”
• “Generate an escalation email about delayed deliverables”

---

🎯 ITIL 4 Consultant

File: itil4-consultant/

Expert ITIL 4 consultant providing context-aware recommendations for IT service management improvement.

When to use:

• Implementing or improving ITIL 4 practices in your organization
• Assessing IT service management maturity
• Optimizing IT operations and service delivery
• Aligning IT services with business objectives
• Designing processes based on ITIL 4 best practices
• Integrating DevOps/Agile with ITIL 4 framework

Core Capabilities:

• ✅ Comprehensive ITIL 4 knowledge (34 practices across 3 categories)
• ✅ Context-aware consulting (department, scenario, industry-specific)
• ✅ Maturity assessment framework (5 levels)
• ✅ Tailored improvement roadmaps
• ✅ Integration with DevOps, Agile, and other frameworks
• ✅ Best practices from extensive knowledge base

Knowledge Base:

• ITIL 4 Foundation concepts (SVS, SVC, 4 Dimensions, 7 Guiding Principles)
• 14 General Management Practices (Architecture, Security, Risk, etc.)
• 17 Service Management Practices (Incident, Change, Release, etc.)
• 3 Technical Management Practices (Deployment, Infrastructure, Development)

Core Workflows:

1. Context Analysis & Scoping - Understand organizational context and needs
2. Current State Assessment - Maturity evaluation with gap analysis
3. Recommendation Development - Tailored, actionable recommendations
4. Roadmap Creation - Phased implementation plan
5. Department/Scenario-Specific Consulting - Context-aware guidance

Scenario Support:

• IT Operations (high incident volume, MTTR improvement)
• Development Teams (change failure reduction, DevOps integration)
• Security Teams (compliance, risk management)
• PMO (portfolio optimization, resource management)
• Infrastructure Teams (cloud migration, platform management)

Deliverables:

• Maturity Assessment Reports
• Gap Analysis with prioritization
• Implementation Roadmaps (phased approach)
• Practice-specific improvement plans with KPIs
• Integration strategies with DevOps/Agile

Key Features:

• 7 Guiding Principles application to all recommendations
• 4 Dimensions holistic consideration
• Quick Wins → Medium-term → Strategic phased approach
• Value-focused consulting (ITIL 4 core principle)
• Best practices from comprehensive knowledge base
• Bilingual: Japanese and English support

---

☁️ Salesforce Expert

File: skill-packages/salesforce-expert.skill

Expert guidance for Salesforce system development and operations management.

When to use:

• Configuring sharing settings, approval processes, validation rules
• Diagnosing access denied issues, approval process errors, missing relationships
• Designing system architecture, data models, object relationships
• Implementing Apex triggers, batch jobs, Lightning Web Components, REST APIs
• Resolving OWD settings, sharing rules, role hierarchy issues
• Addressing governor limit issues, slow queries, large data volume challenges

Core Capabilities:

• ✅ Sharing settings & access control troubleshooting
• ✅ Approval process configuration & debugging
• ✅ Custom development best practices (Apex, LWC)
• ✅ Architecture design & data modeling
• ✅ Governor limit optimization
• ✅ Integration patterns (REST API, Platform Events)

Reference Guides:

• sharing_settings_guide.md - OWD, sharing rules, role hierarchy, manual sharing patterns
• approval_process_guide.md - Single/multi-step approvals, submission methods, testing
• custom_development_patterns.md - Trigger handlers, batch apex, LWC, testing patterns
• architecture_best_practices.md - Data modeling, LDV design, integration, security

Bug Analysis Workflow:

1. Gather Context - Record type, user role, error message, reproduction steps
2. Categorize Issue - Access/permissions, data integrity, process automation, integration, UI/UX
3. Diagnose Root Cause - Systematic troubleshooting with debug queries
4. Propose Solutions - Quick fix, proper solution, long-term improvement
5. Document Fix - Root cause analysis, testing steps, deployment considerations

Development Patterns:

• Trigger Handler pattern with recursion prevention
• Service Layer pattern for reusable business logic
• Batch Apex (standard, iterable, stateful)
• Queueable Apex with chaining
• Lightning Web Components (wire, imperative, pub/sub)
• REST API design (inbound/outbound)

Common Scenarios:

• “User cannot access related Contact after Opportunity reassignment” → OWD/Sharing analysis
• “No applicable approval process found” → Entry criteria and active process validation
• “Contact shows 0 Opportunities despite linkage” → Relationship type analysis
• “Design REST API for external system integration” → REST resource pattern with proper HTTP methods
• “Optimize trigger for 200 records” → Bulkification and governor limit patterns

Best Practices:

• Security: Always use with sharing, check CRUD/FLS, principle of least privilege
• Performance: Bulkify code for 200 records, avoid SOQL in loops, batch DML operations
• Maintainability: Trigger → Handler → Service pattern, single responsibility
• Testing: Test Data Factory, bulk testing, mock callouts, 85%+ coverage

Key Features:

• Systematic Troubleshooting - Step-by-step diagnosis workflows
• Enterprise Patterns - Scalable, maintainable code patterns
• Complete Code Examples - Real-world Apex triggers, handlers, services, LWC
• Architecture Guidance - Data modeling, integration, security, testing
• Bilingual: Japanese and English support

---

🤖 AI Adoption Consultant

File: skill-packages/ai-adoption-consultant.skill

Expert AI/LLM adoption consultant with comprehensive use case knowledge across industries, departments, and scenarios.

When to use:

• Proposing AI/LLM adoption strategies for specific industries (Finance, Healthcare, Retail, Manufacturing, Education)
• Recommending AI solutions for department-specific needs (Sales, HR, Support, Finance, R&D)
• Creating AI implementation plans for different scenarios (Startup, Enterprise, Remote Work, CX, DX)
• Selecting appropriate AI agent types (RAG, Voice, Video Generation, Internal Business Support)
• Analyzing case studies and ROI for AI initiatives

Core Capabilities:

• ✅ Industry-specific AI adoption strategies (5 industries)
• ✅ Department-specific automation & efficiency improvements (5 functions)
• ✅ Scenario-based implementation approaches (5 scenarios)
• ✅ AI agent type recommendations (4 types: RAG, Voice, Video, Business Support)
• ✅ Detailed case studies (6 use cases: Sales Support, Customer Support, Knowledge Search, Project Planning, Competitive Analysis, Strategic Planning)

Knowledge Base:

• Industries: Finance, Healthcare, Retail, Manufacturing, Education
• Functions: Sales & Marketing, HR, Customer Support, Finance & Accounting, R&D
• Scenarios: Startup, Enterprise, Remote Work, Customer Experience, Digital Transformation
• Agent Types: RAG agents, Voice agents, Video generation, Internal business support
• Case Studies: 6 detailed implementation examples with architecture and ROI

Consulting Workflow:

1. Hearing & Current State Assessment - Understand industry, department, company size, specific challenges, existing systems, goals/KPIs, budget, technical skills
2. Problem Analysis & Goal Setting - Identify key issues, define success metrics, organize constraints, identify risks
3. Knowledge Selection & Loading - Load appropriate references (industry + function + scenario + agent type + case studies)
4. AI Adoption Proposal Creation - Executive summary, current analysis, AI utilization proposal (use cases, architecture, data flow), expected effects (quantitative & qualitative), implementation plan (3 phases), investment breakdown, risks & countermeasures, recommendations
5. Follow-up & Q&A - Technical details, alternative options, customization, additional case studies

Proposal Components:

• Executive Summary with ROI
• Current State Analysis
• AI Utilization Proposal (Technology, Use Cases, Architecture, Data Flow)
• Expected Effects (Cost reduction %, Efficiency improvement %, Quality metrics)
• 3-Phase Implementation Plan (Pilot → Full Deployment → Continuous Improvement)
• Investment Breakdown (Initial, Operational, TCO, ROI calculation)
• Risk Assessment & Mitigation
• Best Practices & Recommendations

Example Proposals:

• Retail CX Enhancement: Personalized recommendations, 24/7 chatbot support, review analysis, dynamic pricing → 20% conversion rate increase, 15% cart abandonment reduction, 30% satisfaction improvement
• Startup Sales Efficiency: AI lead generation/scoring, automated proposal creation, personalized sales emails, transcription/analysis → 50% prep time reduction, 25% deal closure rate increase, 2x monthly leads
• Manufacturing Quality Control: Image-based defect detection, sensor data anomaly detection, automated quality reports → 95%+ detection rate, 70% inspection time reduction, 30% quality cost reduction

Key Principles:

• Practice-oriented with specific implementation methods
• Quantified effects (ROI, cost reduction, efficiency improvement)
• Phased approach (Pilot → Deployment → Improvement)
• Risk management (technical, organizational, legal)
• Human-centric (AI augments human capabilities)

Knowledge Resources (27 files, 162KB):

• 5 industry guides
• 5 function guides
• 5 scenario guides
• 4 agent type guides
• 6 detailed case studies
• README with AI adoption trends and success factors

Key Features:

• Comprehensive Knowledge - 27 curated use case documents
• Multi-dimensional Analysis - Industry × Function × Scenario combinations
• Quantified Benefits - All proposals include ROI and cost reduction metrics
• Practical Examples - Real-world case studies with implementation details
• Agent Type Expertise - RAG, Voice, Video, Business Support recommendations
• Bilingual: Japanese and English support

---

☁️ Render CLI Expert

File: skill-packages/render-cli-expert.skill

Expert skill for managing Render cloud platform services via CLI. Supports deployments, log monitoring, SSH connections, PostgreSQL connections, and service management.

When to use:

• Deploying and managing Render services from the terminal
• Monitoring service logs in real-time
• Connecting to PostgreSQL databases via psql
• SSH remote connections to services
• Automating Render operations in CI/CD pipelines
• Listing workspaces and services

Core Capabilities:

• ✅ Service deployment with wait and confirmation options
• ✅ Real-time log monitoring and JSON output
• ✅ PostgreSQL database connections via psql
• ✅ SSH access to paid services
• ✅ CI/CD automation patterns (GitHub Actions, shell scripts)
• ✅ Workspace and service management

Key Commands:

CI/CD Integration:

# Environment variable authentication
export RENDER_API_KEY=rnd_xxx...

# Deploy with wait and auto-confirm
render deploys create srv-abc123 --wait --confirm

Auto-Update Feature:

• Automatic documentation check every month
• Manual update: python3 scripts/render_cli_updater.py
• Force update: python3 scripts/render_cli_updater.py --force

Best Practices:

• Use RENDER_API_KEY environment variable for CI/CD
• Use --wait flag to confirm deployment completion
• Use -o json for automation scripts
• Use --confirm to skip confirmation prompts

Resources:

• scripts/render_cli_updater.py - Auto-update checker script
• references/cli_updates.md - Latest update log
• references/last_check.json - Last check timestamp

---

🔍 Design Implementation Reviewer

File: skill-packages/design-implementation-reviewer.skill

A critical code review skill focused on whether code actually works correctly—not just whether it matches a design document.

When to use:

• Critical code review sessions
• Verifying implementation against requirements
• Finding bugs the design didn’t anticipate
• Reviewing data pipelines and ETL processes

Core Capabilities:

• ✅ Three-layer review framework (Code Quality → Execution Flow → Goal Achievement)
• ✅ Concurrency, idempotency, and transaction boundary checks
• ✅ Resource management and timeout/retry logic verification
• ✅ API contract and backward compatibility analysis
• ✅ Design gap identification

Key Features:

• Always uses ultrathink mode for thorough analysis
• Security review is out of scope (use dedicated security skill)
• Structured output: Review Scope → Findings → Open Questions
• Test Plan required for Critical/High severity findings

---

🎬 Video2Minutes

File: skill-packages/video2minutes.skill

Automatically transcribes video files and generates structured meeting minutes.

When to use:

• Converting meeting recordings to text
• Creating meeting minutes from video content
• Extracting key points and action items from recorded discussions

Core Capabilities:

• ✅ Video transcription using Whisper
• ✅ Automatic meeting minutes generation
• ✅ Key points and action items extraction

---

🔍 Migration Validation Explorer

File: skills/migration-validation-explorer/SKILL.md

Exploratory data-migration validation and QA ideation workflow for CRM migrations. Surfaces hidden data quality risks and generates prioritized QA backlogs with automated testing.

When to use:

• Validating data migration projects (especially CRM migrations)
• Discovering hidden data quality risks
• Generating new validation angles from mapping specs
• Automated data profiling and hypothesis testing
• Exploring validation gaps in migration datasets

Core Capabilities:

• ✅ 4-Perspective Hypothesis Generation (Domain, Tech, Edge Case, Statistical)
• ✅ Priority Scoring (Impact × Probability × Testability)
• ✅ Automated Scripts for profiling, testing, and lens combination
• ✅ Cross-pollination with operators (AND/XOR/SEQ/REQ)
• ✅ QA backlog generation with risk prioritization

4-Perspective Framework:

Automation Scripts:

• scripts/exploratory_profiler.py - Data profiling with null rates, distributions
• scripts/hypothesis_tester.py - Reference integrity, value concentration tests
• scripts/perspective_combiner.py - Lens catalog and random combinations

Workflow:

Example Use Cases:

• “Profile this migration dataset and find potential issues”
• “Generate hypotheses from 4 perspectives for Account data”
• “Test reference integrity between Contacts and Accounts”
• “Create a prioritized QA backlog for go-live”

---

📞 Helpdesk Responder

File: skills/helpdesk-responder/SKILL.md

Generic helpdesk first-response skill for creating KB-based response drafts. Adaptable to any industry or product support context.

When to use:

• Handling customer support tickets or inquiries
• Creating response drafts based on knowledge base articles
• Building a structured helpdesk workflow
• Training support staff on response patterns

Core Capabilities:

• ✅ Auto-detection patterns (error codes, keywords, symptoms)
• ✅ Confidence scoring for response selection (High/Medium/Low)
• ✅ Multi-language response templates (English/Japanese)
• ✅ KB index configuration with JSON schema
• ✅ Escalation workflow with handoff format

Workflow:

Confidence Thresholds:

• High (>=80%): Provide solution directly
• Medium (50-79%): Request additional information
• Low (<50%): Escalate for investigation

Key Components:

• references/kb_schema.json - Complete JSON schema for KB index configuration
• assets/response_templates.md - Ready-to-use response templates (7 scenarios)

Example Use Cases:

• “Create a response draft for ticket #1234”
• “Handle this customer inquiry about error code 30001”
• “Generate a Japanese response for this support request”
• “Set up a KB-based helpdesk workflow for my product”

---

📧 Email Triage Responder

File: skills/email-triage-responder/SKILL.md

Analyze inbox emails to identify action-required items, prioritize by urgency/importance using the Eisenhower Matrix, classify by topic, and generate contextual draft responses.

When to use:

• Triaging a large inbox with many unread emails
• Prioritizing which emails need immediate attention
• Classifying emails by topic or sender type
• Generating draft responses for common email types
• Tracking response status across multiple emails

Core Capabilities:

• ✅ Urgency detection (time-sensitive language, deadlines, escalation markers)
• ✅ Importance scoring (VIP senders, CC patterns, content signals)
• ✅ Topic classification (client, vendor, internal, meeting, FYI)
• ✅ Eisenhower Matrix quadrant assignment (Q1-Q4)
• ✅ Contextual draft response generation (tone/language aware)
• ✅ Response status tracking (pending, draft_ready, sent, delegated)

Workflow:

Priority Matrix:

Key Components:

• scripts/triage_emails.py - Main triage and draft generation script
• references/email-classification.md - Topic taxonomy and urgency markers
• references/response-templates.md - Draft response templates by category

Example Use Cases:

• “Triage my unread emails and show me what needs immediate attention”
• “Generate draft responses for my Q1 and Q2 emails”
• “Classify my inbox by topic and sender type”
• “Track which emails I’ve responded to this week”

---

📥 Inbox Triage Summarizer

File: skills/inbox-triage-summarizer/SKILL.md

Periodic inbox scan that categorizes new emails by project/client, identifies action-required items, and generates prioritized summary with clear action classifications (FYI, requires-response, requires-action, blocked-waiting-on-others).

When to use:

• Running a periodic inbox review (daily, weekly) to understand backlog
• Grouping emails by project or client for focused processing
• Identifying blocked items waiting on external responses
• Generating an executive summary of inbox state for team standup
• Feeding action-required emails to email-triage-responder for response drafting

Core Capabilities:

• ✅ Project/client detection (domain matching, subject keywords, labels)
• ✅ 4-way action classification (FYI, requires-response, requires-action, blocked-waiting)
• ✅ Thread correlation and staleness tracking
• ✅ Urgency scoring with deadline detection
• ✅ Export format compatible with email-triage-responder

Classification Categories:

Key Components:

• scripts/triage_inbox.py - Main inbox triage and summarization script
• references/email-categorization-guide.md - Classification criteria and rules
• references/project-mapping.yaml - Project/client detection configuration

Example Use Cases:

• “Scan my inbox and group emails by project”
• “Show me which threads are blocked waiting on responses”
• “Generate a summary of action items for today’s standup”
• “Export action-required emails for response drafting”

---

📬 Email Thread Summarizer

File: skills/email-thread-summarizer/SKILL.md

Analyze email threads to extract current status, pending actions, timeline of events, and identify stale/outdated information. Supports both Gmail (via gogcli) and Outlook email exports.

When to use:

• Summarizing long email conversations to understand current state
• Extracting outstanding action items from email exchanges
• Building a timeline of key events and decisions from email threads
• Detecting if a previously cached thread summary is outdated
• Analyzing thread status (resolved, pending, stale, escalated)

Core Capabilities:

• Multi-format parsing (Gmail JSON, Outlook/Graph API, EML, MBOX)
• Participant role detection (initiator, responder, CC)
• Action item extraction with deadline detection
• Decision identification and tracking
• Superseded information detection
• Thread status classification (active, resolved, stale, escalated, awaiting_response)
• Staleness detection for cached summaries

Workflow:

Status Classification:

Key Components:

• scripts/analyze_thread.py - Parse thread and extract structured data
• scripts/generate_summary.py - Generate markdown summary report
• scripts/check_staleness.py - Compare cached vs current state
• references/thread_analysis_patterns.md - Action item, decision, and status patterns

Example Use Cases:

• “Summarize this email thread and tell me what’s still pending”
• “What decisions were made in this conversation?”
• “Is my summary from last week still accurate?”
• “Who committed to what actions in this thread?”

---

🎬 FFmpeg Expert

File: skill-packages/ffmpeg-expert.skill

A comprehensive FFmpeg CLI skill for multimedia processing - video/audio conversion, editing, and optimization.

When to use:

• Converting video formats (MP4, WebM, MKV, MOV, AVI)
• Transcoding codecs (H.264, H.265/HEVC, VP9, AV1, ProRes)
• Trimming, cutting, or merging videos
• Extracting or converting audio (MP3, AAC, FLAC, WAV)
• Creating GIFs from video
• Generating thumbnails
• Preparing for streaming (HLS/DASH)
• Using hardware acceleration (NVENC, QSV, VideoToolbox)
• Applying filters (blur, color correction, subtitles, watermarks)

Key Components:

• references/quick_reference.md - Copy-paste ready commands
• references/codec_guide.md - Codec selection guide (H.264, H.265, VP9, AV1)
• references/filter_reference.md - Filter syntax and examples
• references/troubleshooting.md - Common errors and solutions
• scripts/ffprobe_analyzer.py - Media analyzer with encoding recommendations

Example Use Cases:

• “Convert this video to MP4”
• “Extract audio from video”
• “Create a GIF from 1:00 to 1:10”
• “Compress with H.265”
• “Add a watermark”
• “Prepare for HLS streaming”

---

📊 Lean Six Sigma Consultant

File: skill-packages/lean-six-sigma-consultant.skill

Comprehensive Lean Six Sigma consulting skill supporting all belt levels (White Belt to Master Black Belt).

When to use:

• Leading or supporting process improvement projects
• DMAIC/DMADV methodology guidance
• Lean waste elimination (VSM, 8 Wastes/DOWNTIME, 5S)
• Statistical analysis (process capability Cp/Cpk, control charts, hypothesis testing)
• Root cause analysis (5 Whys, Fishbone)
• Six Sigma training and certification preparation

Key Components:

• references/methodology/ - DMAIC, DMADV, Lean principles
• references/tools-by-phase/ - Define, Measure, Analyze, Improve, Control
• references/statistics/ - Control charts, hypothesis testing, process capability
• references/lean-tools/ - 5S, VSM, Kaizen, 8 Wastes
• scripts/sigma_calculator.py - Calculate sigma level and DPMO
• scripts/control_chart_analysis.py - Control chart generation and analysis
• scripts/process_capability.py - Cp/Cpk calculation

Example Use Cases:

• “Guide me through a DMAIC project to reduce defects”
• “Calculate the Cpk for this process data”
• “Create a value stream map for our order fulfillment”
• “What control chart should I use for defect counts?”
• “Explain the 8 wastes in our service process”

---

🦆 DuckDB Expert

File: skill-packages/duckdb-expert.skill

Expert skill for large-scale data analysis using DuckDB - the embedded OLAP database.

When to use:

• Querying large CSV, Parquet, JSON files directly
• Analyzing datasets that don’t fit in memory
• Building ETL pipelines with SQL
• Integrating with pandas/Polars
• Building file-based data warehouses
• Optimizing complex analytical queries

Key Components:

• references/duckdb_functions_reference.md - Function reference
• references/file_formats_guide.md - CSV, Parquet, JSON handling
• references/performance_tuning_guide.md - Query optimization
• scripts/duckdb_analyzer.py - Data analysis automation
• scripts/etl_pipeline.py - ETL pipeline utilities

Example Use Cases:

• “Analyze this 10GB Parquet file”
• “Query multiple CSV files with a wildcard pattern”
• “Optimize this slow analytical query”
• “Build an ETL pipeline for data transformation”

---

🔍 Critical Code Reviewer

File: skill-packages/critical-code-reviewer.skill

Self-contained multi-persona code review skill using four expert perspectives for thorough quality assessment. Agent prompts are embedded in the skill package — no external dependencies required.

Reviewer Personas:

When to use:

• Reviewing source code from multiple expert perspectives
• Finding design flaws, testability issues, and runtime bugs
• Assessing code quality and maintainability
• Python/JavaScript gets Tier 1 deep checks; IaC/Config files also supported

Key Components:

• references/agents/*.md - 4 embedded persona prompts (self-contained)
• references/code_smell_patterns.md - Code smells and anti-patterns
• references/review_framework.md - Critical analysis framework
• references/severity_criteria.md - Severity judgment criteria
• references/file_type_classification.md - File type auto-classification with content sniffing
• references/scale_strategy.md - Large codebase hotspot extraction strategy

---

📄 Critical Document Reviewer

File: skill-packages/critical-document-reviewer.skill

Self-contained multi-persona document review skill for rigorous validation of claims and evidence. Agent prompts are embedded in the skill package — no external dependencies required.

Reviewer Personas (up to 6, selected by document type):

When to use:

• Reviewing design documents, analysis reports, or proposals
• Validating claims have proper evidence
• Detecting unsupported claims and logical gaps
• Finding missing traceability to requirements
• Security/operations review of technical documents

Key Components:

• references/agents/*.md - 6 embedded persona prompts (self-contained)
• references/severity_criteria.md - Document review severity criteria
• references/red_flag_patterns.md - Dangerous expression patterns (context-sensitive)
• references/persona_selection_matrix.md - Document type to persona mapping
• references/scale_strategy.md - Large document review strategy

---

🎨 Design Thinking

File: skill-packages/design-thinking.skill

Human-centered innovation and problem-solving skill based on Stanford d.school / IDEO methodology.

5 Phases:

EMPATHIZE → DEFINE → IDEATE → PROTOTYPE → TEST
   共感    →  定義  →  発想  →   試作   → 検証

When to use:

• New service/product planning
• Customer experience (CX/UX) improvement
• Problem redefinition (“What problem should we really solve?”)
• Innovation creation and brainstorming facilitation

Key Components:

• references/empathize_methods.md - User research methods
• references/define_methods.md - Problem framing (HMW, POV)
• references/ideate_methods.md - Brainstorming techniques
• references/prototype_methods.md - Rapid prototyping
• references/test_methods.md - User testing methods
• Templates: Persona, Empathy Map, Journey Map, Ideation Canvas

---

📊 DAMA-DMBOK

File: skill-packages/dama-dmbok.skill

Data management skill based on DAMA-DMBOK (Data Management Body of Knowledge).

11 Knowledge Areas:

• Data Governance (central/governing)
• Data Quality, Data Architecture, Data Security
• Reference & Master Data Management (MDM)
• Data Integration, Metadata Management
• Data Modeling, DWH & BI, Data Storage
• Document & Content Management

When to use:

• Data strategy development
• Data governance framework implementation
• Data quality improvement
• Data catalog creation
• MDM implementation
• Data migration projects
• Data maturity assessment

Key Components:

• references/data_governance.md - Governance framework
• references/data_quality.md - Quality dimensions and improvement
• references/metadata_management.md - Metadata standards
• references/master_data_management.md - MDM patterns
• Templates: Governance Charter, Maturity Assessment, Data Catalog

---

🔧 TDD Developer

File: skill-packages/tdd-developer.skill

Test-Driven Development methodology skill for writing tests before implementation.

TDD Cycle:

RED → GREEN → REFACTOR
(Write failing test → Make it pass → Improve code)

When to use:

• User requests TDD or test-first development
• Building new features requiring high reliability
• Refactoring legacy code with test coverage
• Implementing complex business logic
• Learning TDD methodology

Key Components:

• references/tdd_methodology.md - TDD principles and workflow
• references/test_patterns.md - Common testing patterns

---

📈 QA Bug Analyzer

File: skill-packages/qa-bug-analyzer.skill

Bug ticket analysis skill for quality trends, hotspot detection, and improvement recommendations.

When to use:

• Analyzing bug ticket data (CSV, Excel, JSON, Markdown)
• Quality assessment for project reviews or release readiness
• Trend analysis over time
• Generating stakeholder reports
• Identifying improvement priorities

Key Components:

• scripts/analyze_bugs.py - Automated bug analysis
• references/analysis_methodology.md - Analysis approach
• references/quality_metrics_guide.md - Quality metrics
• assets/report_template.md - Report format

Example Use Cases:

• “Analyze bug trends from this CSV”
• “Generate a quality report for project management”
• “Which modules have the most bugs?”
• “What areas should we focus on for improvement?”

---

📑 UAT Testcase Generator

File: skill-packages/uat-testcase-generator.skill

Generates UAT (User Acceptance Testing) test cases in Excel format for Salesforce CRM projects.

When to use:

• Creating standardized UAT test cases for Salesforce projects
• Generating Excel files with summary and detailed test cases
• Documenting test scenarios, preconditions, steps, and expected results

---

🔄 Docling Converter

File: skill-packages/docling-converter.skill

Document conversion skill using the docling CLI - convert any document to Markdown and other formats.

Supported Input: PDF, DOCX, PPTX, HTML, Images, Excel Supported Output: Markdown, JSON, YAML, HTML, Text

When to use:

• Converting documents to Markdown
• Extracting text from PDFs
• Processing scanned documents with OCR
• Converting office documents

Example:

/docling-converter document.pdf --to md --ocr-lang ja

---

☁️ AWS CLI Expert

File: skill-packages/aws-cli-expert.skill

AWS CLI expert skill for cloud infrastructure management and operations.

Key Components:

• references/aws_cli_essentials.md - Core CLI usage
• references/iam_guide.md - IAM and security
• references/security_best_practices.md - Security guidelines

---

☁️ gogcli Expert

File: skill-packages/gogcli-expert.skill

Expert skill for gogcli (openclaw/gogcli), a Go-based CLI tool for managing 25+ Google Workspace and related services from the terminal. Tracks gogcli v0.27.0.

When to use:

• Managing Gmail (search, send, labels, settings, tracking)
• Calendar operations (events, conflicts, free/busy, focus-time, out-of-office, recurring)
• Drive file operations (ls, upload, download, share, audit, bulk permissions)
• Sheets data reading/writing (A1 notation, append, formatting, charts, tables)
• Docs/Slides full editing (create, find-replace, create-from-markdown, export PDF/DOCX/PPTX)
• Tasks management (create, complete, recurring)
• Workspace admin (Groups, Directory API Admin, Classroom, People, Contacts)
• New services (Forms, Meet, Maps, YouTube, Photos, Sites, Analytics, Search Console, Apps Script, Zoom)
• Setting up OAuth2 / service account authentication
• Multi-account and multi-client configuration

Core Capabilities:

• 25+ services: Gmail, Calendar, Chat, Drive, Sheets, Docs, Slides, Tasks, Keep, Contacts, People, Groups, Admin, Classroom, Forms, Meet, Maps, YouTube, Photos, Sites, Analytics, Search Console, Apps Script, Zoom (plus backup / batch / mcp)
• OAuth2 + Service Account authentication with scope control
• Multi-account management with aliases and domain mapping
• --json / --plain output for pipeline integration
• Agent safety: --gmail-no-send, --dry-run, --enable-commands sandboxing, built-in mcp server

Key Components:

• references/quick_reference.md - All-services command cheat sheet
• references/communication_services.md - Gmail/Calendar/Chat detailed guide
• references/productivity_services.md - Drive/Sheets/Docs/Slides/Tasks/Keep detailed guide
• references/workspace_admin_services.md - Groups/Admin/Classroom/People + new services guide
• references/troubleshooting.md - Comprehensive troubleshooting guide

---

☁️ Salesforce CLI Expert

File: skill-packages/salesforce-cli-expert.skill

Salesforce CLI (sf/sfdx) expert skill for Salesforce development and administration.

When to use:

• Authenticating to orgs
• Querying data with SOQL
• Retrieving metadata (profiles, permission sets, security settings)
• Deploying configuration changes
• Automating security audits

---

🤖 Codex Reviewer

File: skill-packages/codex-reviewer.skill

Uses OpenAI Codex CLI to review documents and code with GPT-6.1-Sol model (xhigh reasoning).

When to use:

• Code review requiring deep analysis
• Document review
• Design document review
• Test plan review

---

📊 Markdown to PDF

File: skill-packages/markdown-to-pdf.skill

Converts Markdown documents to professional PDF with two rendering modes: (1) fpdf2 mode for business documents with cover pages, themed styling, and styled tables, (2) Playwright mode for technical documentation with Mermaid diagram support.

When to use:

• Creating professional business PDFs (estimates, proposals, reports)
• Converting Markdown with cover pages and themed headers/footers
• Converting technical documentation with Mermaid diagrams
• Extracting Mermaid diagrams as images

Key Components:

• scripts/markdown_to_fpdf.py - Professional PDF (fpdf2, cover pages, themes, tables)
• scripts/markdown_to_pdf.py - Mermaid-aware PDF (Playwright/HTML/CSS)
• scripts/themes.py - Theme definitions and CJK font discovery
• scripts/mermaid_to_image.py - Mermaid to PNG/SVG

---

📋 Meeting Asset Preparer

File: skills/meeting-asset-preparer/

Prepares comprehensive meeting assets including agendas, reference materials, decision logs, and follow-up action items. Integrates with project context to pull relevant documents, estimates, and implementation differences. Supports bilingual (Japanese/English) output for cross-regional meetings.

When to use:

• Preparing assets for an upcoming project meeting or status review
• Creating bilingual meeting documentation for cross-regional teams
• Compiling reference materials from estimates, specs, and implementation docs
• Generating structured agendas with time allocations
• Setting up decision log and action item tracking templates

Key Components:

• scripts/prepare_meeting.py - Main CLI for meeting asset preparation (init, compile-refs, generate-agenda, create-decision-log, create-action-items, package)
• references/meeting-best-practices.md - Best practices for effective meeting preparation
• assets/agenda_template.md - Bilingual agenda template
• assets/decision_log_template.md - Bilingual decision log template
• assets/action_items_template.md - Bilingual action items template

Key Features:

• Bilingual support (Japanese/English) for cross-regional teams
• Context integration from project documents (estimates, specs, prior notes)
• Structured templates for agendas, decision logs, and action items
• Time allocation in agendas for focused meetings
• Meeting package generation with index document

---

📝 Meeting Minutes Reviewer

File: skills/meeting-minutes-reviewer/

Reviews meeting minutes documents for completeness, action item clarity, decision documentation, and consistency with source materials. Generates structured feedback with specific improvement suggestions and quality scores across 5 dimensions.

When to use:

• After drafting meeting minutes and before distribution
• When reviewing minutes created by others for quality assurance
• When validating that minutes accurately reflect source materials (hearing sheets, transcripts)
• When ensuring action items meet trackability standards
• When preparing minutes for formal project documentation or audit trails

Key Components:

• scripts/review_minutes.py - Main review script with 5-dimension quality analysis
• references/review-criteria.md - Detailed scoring criteria and quality standards
• references/meeting-minutes-checklist.md - Complete checklist for meeting minutes

Key Features:

• 5-dimension scoring: Completeness (25%), Action Items (25%), Decisions (20%), Consistency (15%), Clarity (15%)
• Action item validation: owner, deadline, description completeness
• Decision documentation check: context, rationale, alternatives
• Source material consistency verification (hearing sheets, agendas)
• Vague language detection and clarity analysis
• JSON and Markdown report output formats

---

📝 Meeting Minutes Writer

File: skills/meeting-minutes-writer/

Generates strategic-consultant-grade meeting minutes from transcripts or notes, then runs a self-review loop (max 3 iterations) that checks for internal contradictions, action-item omissions, speaker-name errors, and date/day-of-week mistakes before reporting completion.

When to use:

• Converting raw meeting transcripts or notes into structured minutes
• Producing executive-readable minutes (3-minute readability test)
• Any time minutes must include verified dates and complete action items
• When you need quality-gated output (zero findings or 3 iterations) before sharing

Key Components:

• references/output_format.md - Canonical minutes structure, inference rules, ambiguity markers
• references/self_review_checklist.md - 5 Mandatory Checks with severity model and iteration logic
• assets/minutes_template_en.md - Blank meeting minutes template (English)
• assets/minutes_template_ja.md - 議事録テンプレート（日本語）
• assets/findings_report_template.md - Per-iteration findings report layout (bilingual EN + JA)

Key Features:

• 2-phase workflow: ultrathink Generation → Self-Review Loop (max 3 iterations)
• 5 Mandatory Checks: Internal Contradictions, Consistency, Action-Item Omissions, Speaker-Name Errors, Date/Day-of-Week Errors
• MANDATORY date verification via python3 -c "import datetime; ..." (no memory-based dates)
• Severity model: HIGH (blocks completion) / MEDIUM / LOW
• Completion report surfaces remaining HIGH findings and * To be confirmed items
• Complements meeting-minutes-reviewer (review-only) and video2minutes (transcribe-then-write)

---

🇯🇵 Japanese Enterprise Document Formatter

File: skills/japanese-enterprise-doc-formatter/

Formats documents for Japanese enterprise approval workflows including ringi (稟議), purchase requests (購入申請), and internal proposals. Handles bilingual requirements, proper keigo levels, required approval sections, and corporate template compliance.

When to use:

• Creating or formatting 稟議書 (ringi) approval documents
• Preparing 購入申請書 (purchase request) forms
• Drafting 提案書 (internal proposals) for management
• Converting informal notes into formal Japanese business documents
• Ensuring proper keigo levels (尊敬語, 謙譲語, 丁寧語) in documents
• Creating bilingual (Japanese/English) enterprise documents

Key Components:

• scripts/format_document.py - Main document formatting tool with multiple output formats
• scripts/transform_keigo.py - Keigo level transformation (highest/upper/standard/basic)
• scripts/validate_sections.py - Section validation against document type requirements
• references/document_types.md - Document type specifications (稟議書, 購入申請書, 提案書, 報告書, 依頼書)
• references/keigo_guide.md - Keigo usage guide with transformation rules
• references/section_templates.md - Ready-to-use templates for each document type

Key Features:

• 5 document types: 稟議書, 購入申請書, 提案書, 報告書, 依頼書
• 4 keigo levels: 最上級 (executive), 上級 (upper), 標準 (standard), 基本 (basic)
• Section validation with completeness scoring
• Bilingual output with English summaries
• Approval section generation based on document type

---

🎤 Fujisoft Presentation Creator

File: skill-packages/fujisoft-presentation-creator.skill

Creates professional presentations following FUJISOFT America’s slide template standards.

When to use:

• Creating business presentations
• Proposal materials
• MARP-format Markdown presentations

---

📝 Purchase Request Generator

File: skills/purchase-request-generator/

Generates structured IT/hardware purchase request documents from informal requirements. Transforms product details, pricing, and justification notes into formal approval documents including cost-benefit analysis, ROI justification, vendor comparison matrices, and MARP presentation slides suitable for management approval workflows.

When to use:

• Creating formal purchase requests from informal product/price specifications
• Generating cost-benefit analysis for IT equipment or software purchases
• Building ROI justification documents for budget approvals
• Comparing multiple vendors for procurement decisions
• Preparing MARP presentation slides for purchase approval meetings

Key Components:

• scripts/generate_purchase_request.py - Generate formal purchase request documents
• scripts/generate_cba.py - Generate cost-benefit analysis with ROI/NPV calculations
• scripts/generate_vendor_comparison.py - Create vendor comparison matrices
• scripts/generate_marp_slides.py - Generate MARP presentation slides
• references/purchase_request_guide.md - Best practices for purchase justifications

Key Features:

• ROI, NPV, and payback period calculations
• Weighted vendor comparison matrices
• MARP-formatted presentation slides
• Sensitivity analysis and risk scenarios
• Professional document formatting for management approval

---

📊 Office Script Expert

File: skills/office-script-expert/

Office Scripts（Excel Online / Microsoft 365）開発の専門スキル。プラットフォーム制約、ExcelScript APIパターン、テスト戦略、本番開発で発見された13のバグパターンをカバー。

When to use:

• Office Scripts（TypeScript）でExcel Onlineの自動化スクリプトを開発する
• ExcelScript APIのパターンや制約を確認する
• Office Scripts特有のバグパターンを回避する
• lib/抽出 + インライン化のアーキテクチャでテスト可能なコードを書く

Key Features:

• 6つの重要プラットフォーム制約（P1-P6）: import不可、外部ライブラリ不可、120秒タイムアウト等
• lib/抽出 + Vitestによるテスト戦略
• 13の実運用バグパターンと回避策
• ExcelScript API パターン集（シート読み書き、CSV解析、保護/解除、日付変換等）
• 実装前チェックリスト

Reference Guides:

• excel_api_patterns.md - ExcelScript API の9つの主要パターン
• common_bug_patterns.md - 13の実運用バグパターン
• platform_limitations.md - 6つの重要プラットフォーム制約
• testing_strategy.md - lib/抽出 + Vitest テスト戦略

Assets:

• implementation_checklist.md - デプロイ前の実装チェックリスト

---

🌐 Network Diagnostics

File: skills/network-diagnostics/

ネットワーク品質を総合的に診断し、ボトルネックの特定と根本原因の深堀りまで行うスキル。OS標準ツールのみ使用（外部依存なし）、macOSおよびLinux対応。

When to use:

• ネットワーク品質の総合診断
• 接続の遅延・速度低下の原因調査
• レイテンシ・帯域幅・ジッターの測定
• ネットワークヘルスレポートの生成

Key Features:

• 3-Phase ワークフロー: Collect → Analyze & Report → Deep-Dive
• 接続種別（Ethernet/Wi-Fi）対応の品質閾値判定
• 複数CDNによる速度テスト（Cloudflare, OVH, Hetzner）
• HTTP接続タイミング分解（DNS/TCP/TLS/TTFB）
• macOS/Linux クロスプラットフォーム対応（iproute2 + net-tools fallback）

Scripts:

• network_diagnostics.py - データ収集スクリプト（JSON出力、CLI対応）

Reference Guides:

• network_quality_thresholds.md - 品質閾値定義（接続種別×GOOD/WARNING/CRITICAL）
• deep_dive_procedures.md - 6カテゴリの深堀り調査手順

Assets:

• network_report_template.md - 日本語レポートテンプレート（罫線テーブル）

---

🌐 Network Incident Analyzer

File: skills/network-incident-analyzer/

Analyzes network device logs to identify connectivity issues, latency problems, and outages. Automatically correlates timestamps across timezones, detects anomaly patterns during specified time windows, and generates incident reports with root cause hypotheses.

When to use:

• Troubleshooting network connectivity issues using device logs
• Analyzing multiple log files from different network devices
• Investigating service outages with logs spanning multiple timezones
• Identifying patterns in network latency or packet loss
• Generating incident reports with root cause analysis

Key Features:

• Multi-format log parsing (Cisco IOS, JunOS, Palo Alto, F5, Syslog, JSON)
• Automatic timestamp normalization to UTC across timezones
• Anomaly detection (connection failure spikes, interface flapping, error rate spikes)
• Cross-device event correlation with configurable time windows
• Root cause hypothesis generation with confidence scoring
• JSON and Markdown report generation

Scripts:

• analyze_network_logs.py - Main analysis script with CLI interface

Reference Guides:

• log-formats.md - Supported log formats and parsing patterns
• anomaly-patterns.md - Anomaly detection patterns and thresholds

---

🔗 Multi-File Log Correlator

File: skills/multi-file-log-correlator/

Correlates events across multiple log files from different sources, systems, or time periods. Builds unified timelines, identifies causal relationships between events, and highlights anomalies that span multiple data sources. Supports timezone normalization and gap detection.

When to use:

• Investigating incidents involving multiple services (frontend, API, database, cache)
• Analyzing distributed system failures across microservices
• Building event timelines from logs with different timestamp formats
• Correlating infrastructure logs with application logs
• Detecting timing anomalies between system components
• Tracing request flows across multiple services

Key Features:

• Unified timeline construction from heterogeneous log sources
• Correlation ID tracking (request_id, trace_id, transaction_id, UUID)
• Temporal proximity correlation when explicit IDs are absent
• Timezone normalization (supports IANA timezone names)
• Gap detection with configurable thresholds
• Anomaly detection (timing anomalies, error bursts)
• JSON and Markdown report generation

Scripts:

• correlate_logs.py - Main correlation engine with CLI interface

Reference Guides:

• correlation_methodology.md - Correlation techniques and best practices
• timestamp_formats.md - Common log timestamp formats reference

---

🌊 Streamlit Expert

File: skills/streamlit-expert/

Streamlit Webアプリケーション開発の専門スキル。v1.42〜v1.52+の最新機能に対応。

When to use:

• Streamlitアプリの新規構築
• OIDC認証（Google, Microsoft, Okta, Auth0）の実装
• データ可視化ダッシュボードの作成
• パフォーマンス最適化（キャッシュ、セッション管理）

Key Features:

• 認証: st.login() / st.logout() / st.user によるネイティブOIDC
• 可視化: Plotly, Altair, ネイティブチャートの最適選択
• シークレット管理: st.secrets によるセキュアな資格情報管理
• パフォーマンス: キャッシュ戦略、大規模データセット処理
• モダン機能: カスタムテーマ、マルチページアプリ、Custom Components v2

---

✍️ AI Text Humanizer

File: skills/ai-text-humanizer/

AI（LLM）生成テキストの「AI臭」を検出・診断し、人間らしい文章にリライトするスキル。

When to use:

• AI生成テキストの品質チェック・改善
• 「AIっぽい」文章を人間らしくリライトしたいとき
• テキストがAI生成かどうかの簡易判定

6つの検出パターン:

3つの人間化技法:

1. バランスを崩す — 立場を取る、強い断定を使う
2. 客観を崩す — 主観・経験・判断を入れる
3. 論理を崩す — 完璧な構造を壊す、自然な脱線を入れる

Score Interpretation:

Note: 検出スクリプト (detect_ai_patterns.py) は日本語テキスト専用。英語テキストの場合はClaude自身が references/ を参照して分析・リライトする。

🔍 Dual Axis Skill Reviewer

File: skills/dual-axis-skill-reviewer/

スキルの品質を二軸（Auto + LLM）で定量評価するレビュースキル。決定論的なコードベースチェック（構造、スクリプト、テスト、実行安全性）とLLMによる深層レビューを組み合わせ、再現性のあるスコアリングを提供する。

When to use:

• skills/*/SKILL.md の品質スコアリングが必要なとき
• マージゲートとしてスコア閾値（例: 90点以上）を設定したいとき
• 低スコアスキルの具体的改善項目を取得したいとき
• 別プロジェクトのスキルをクロスプロジェクトでレビューしたいとき

Key Features:

• Auto軸: メタデータ(20)、ワークフロー網羅(25)、実行安全性(25)、成果物(10)、テスト健全性(20)の5次元スコアリング
• LLM軸: スクリプトの正確性、リスク、保守性をJSON形式で評価
• 重み付き統合スコア（--auto-weight / --llm-weightで調整可能）
• --all オプションで全スキル一括レビュー＆サマリテーブル出力
• --project-root によるクロスプロジェクト対応
• knowledge_only スキルの自動判定（スクリプト不在でも不当減点を回避）
• 90点未満で改善項目を自動生成

---

📝 Business Plan Creator

File: skills/business-plan-creator/

事業計画書を体系的に作成するスキル。新規事業、既存事業拡大、スタートアップのピッチ資料、社内新規プロジェクト提案など、あらゆる事業計画のドキュメントを構造化して作成する。

When to use:

• 新規事業の事業計画書を作成したいとき
• 投資家向けピッチ資料や銀行融資申請書が必要なとき
• 社内稟議・新規事業提案書を作成したいとき
• 収支シミュレーションや市場分析を含む計画書が必要なとき

Key Features:

• 5フェーズワークフロー（ヒアリング → 分析 → 戦略設計 → 数値計画 → ドキュメント化）
• 7つの分析フレームワーク（TAM/SAM/SOM、PEST、5フォース、SWOT、BMC、リーンキャンバス、バリューチェーン）
• 財務モデリング（ユニットエコノミクス、3シナリオ分析、P/L・CF計画）
• 業種別テンプレート（SaaS、EC/D2C、コンサル、飲食、製造、AI、不動産）
• 標準構成（12セクション）と簡易構成（7セクション）の2パターン
• 日本語・英語の両言語対応

---

🔍 Audit Document Checker

File: skills/audit-doc-checker/

Review audit-related documents against 12 quality categories and produce a structured quality score (0-100) with severity-rated findings.

When to use:

• Reviewing control design documents before submission to external auditors
• Checking quality of bottleneck analyses or risk assessment reports
• Verifying consistency across audit documentation (terminology, currency, accounting standards)
• Pre-publication quality gate for audit-context documents

Key Features:

• 12 check categories (terminology, currency, accounting standards, control logic, assertions, SoD, materiality, etc.)
• Severity-based scoring (High -5, Medium -3, Low -1 per finding)
• Document-type weighting (control design, bottleneck analysis, requirements, process inventory)
• 4-tier quality assessment (90+ High Quality, 70-89 Improvement Recommended, 50-69 Revision Required, <50 Critical Risk)
• Integration with audit-control-designer for generate→review workflow

---

🏗️ Audit Control Designer

File: skills/audit-control-designer/

Generate audit-ready internal control design documents from As-Is business process inventories.

When to use:

• Building internal controls from As-Is process inventories
• Designing controls for new audit engagements (SOX, J-SOX, PCAOB)
• Preparing for initial audit readiness assessment
• Strengthening existing control frameworks

Key Features:

• 5 business process patterns (AP, Inventory, COGS, Returns, Price Management)
• 8 control templates (T-AP-01/02, T-INV-01/02, T-CO-01, T-VAL-01, T-CALC-01/02)
• 5 audit assertion mapping (C/A/V/CO/E) with coverage assessment
• 5 SoD pair analysis with compensating controls for small organizations
• 8 KPI definitions (K01-K08) with baseline/target guidelines
• Materiality framework (Overall, Performance, Clearly Trivial)
• Accounting standards reference (US GAAP / IFRS / J-GAAP differences)
• Short-term (M+1~M+4) and medium-term (M+6~M+18) roadmap generation
• Industry customization (F&B, Retail, Manufacturing, Services)

---

📄 Multi-Format Document Optimizer

File: skill-packages/multi-format-document-optimizer.skill

Unified document optimization skill that chains docling-converter, imagemagick-expert, and markdown-to-pdf to create comprehensive document processing pipelines.

When to use:

• Converting PDF/DOCX/PPTX to optimized PDF with compressed images
• Creating web-ready documents with WebP images from source documents
• Generating print-ready PDFs with high-DPI images from mixed sources
• Batch processing multiple documents through consistent optimization pipeline
• Extracting, optimizing, and re-embedding images from existing documents
• Converting scanned documents with OCR and image enhancement

Core Capabilities:

• ✅ Automatic format detection and pipeline routing
• ✅ Quality presets (web, print, archive, minimal, custom)
• ✅ Image optimization with configurable quality, DPI, and format
• ✅ PDF image extraction and re-embedding
• ✅ Batch processing with parallel workers
• ✅ Document verification and quality assessment

Key Features:

Automated Scripts:

• document_optimizer.py - Main CLI tool with analyze/convert/batch/optimize-images/verify commands

Reference Guides:

• pipeline_guide.md - Detailed pipeline configurations and customization options
• image_optimization_guide.md - Image optimization strategies and quality settings

Quality Presets:

Example Use Cases:

• “Convert this PPTX to PDF and optimize images for web”
• “Make this scanned PDF smaller while keeping it readable”
• “Batch process all documents in this folder for web publishing”
• “Optimize this PDF’s embedded images”

---

Installation

Installing a Skill

1. Download the desired .skill file from skill-packages/
2. For Claude Desktop: Settings → Capabilities → Upload the .skill file
3. For Claude Code CLI: Copy the skill folder to ~/.claude/skills/
4. The skill will be available for use immediately

Skill Structure

Each skill follows this standard structure:

skill-name/
├── SKILL.md              # Main skill documentation and workflow
├── scripts/              # Automated tools and utilities
├── references/           # Methodology guides and best practices
└── assets/               # Templates and boilerplate files

Development

Creating New Skills

This repository uses the skill-creator tool from Anthropic’s agent-skills marketplace.

Prerequisites:

• Claude Code installed
• example-skills:skill-creator plugin available

Steps to create a new skill:

1. Initialize skill structure:

   python /path/to/skill-creator/scripts/init_skill.py  --path ./
   

2. Develop skill contents:

   • Edit SKILL.md with workflow and instructions
   • Add executable scripts to scripts/
   • Add reference documentation to references/
   • Add templates to assets/

3. Package the skill:

   python /path/to/skill-creator/scripts/package_skill.py ./skill-name ./
   

Skill Development Best Practices

SKILL.md Guidelines:

• Use imperative/infinitive form (verb-first instructions)
• Include clear “When to Use” section with specific scenarios
• Provide concrete examples and workflows
• Reference bundled resources appropriately

Resource Organization:

• scripts/ - Executable code that can run without loading to context
• references/ - Documentation loaded on-demand to inform decisions
• assets/ - Files used in output (templates, boilerplate)

Quality Standards:

• Follow progressive disclosure (metadata → SKILL.md → resources)
• Document domain knowledge and procedural steps
• Provide decision trees for complex workflows
• Include troubleshooting sections

Contributing

Contributions are welcome! To contribute a new skill:

1. Fork this repository
2. Create your skill following the structure above
3. Test thoroughly with various use cases
4. Submit a pull request with:
   • Skill .skill file
   • Update to this README
   • Example usage scenarios

License

[Specify your license here]

Contact

For questions or suggestions, please open an issue in this repository.

---

Detailed Skill Descriptions

Data Scientist Skill - Deep Dive

Analysis Workflow

The data-scientist skill follows a systematic 7-step approach:

1. Problem Definition & Scoping

   • Translate business problems into data science problems
   • Define success criteria and constraints

2. Data Understanding (EDA)

   • Automated quality assessment
   • Distribution and relationship analysis
   • Correlation exploration

3. Data Preparation & Feature Engineering

   • Mathematical transformations
   • Categorical encoding
   • Interaction features
   • Time-based features
   • Domain-specific features

4. Model Selection & Training

   • Automated model comparison
   • Hyperparameter tuning
   • Cross-validation

5. Model Evaluation & Interpretation

   • Multiple metrics assessment
   • Feature importance analysis
   • SHAP value interpretation
   • Residual analysis

6. Time Series Analysis (when applicable)

   • Stationarity testing
   • Seasonal decomposition
   • Forecasting with confidence intervals

7. Insights & Recommendations

   • Translate findings to business impact
   • Generate professional reports
   • Provide actionable recommendations

Quick Start Examples

Example 1: Customer Churn Prediction

# Step 1: Automated EDA
python scripts/auto_eda.py customer_data.csv --target churned --output eda_results/

# Step 2: Model Comparison
python scripts/model_comparison.py customer_data.csv churned --problem-type classification --output models/

# Step 3: Review results and generate insights

Example 2: Sales Forecasting

# Time series analysis and forecasting
python scripts/timeseries_analysis.py sales.csv revenue --date-col date --forecast-periods 30 --output forecast/

Example 3: Regression Analysis

# Step 1: EDA
python scripts/auto_eda.py housing_data.csv --target price --output eda/

# Step 2: Model comparison
python scripts/model_comparison.py housing_data.csv price --problem-type regression --output models/

Advanced Features

Handling Imbalanced Data:

• SMOTE oversampling
• Class weight adjustment
• Appropriate metric selection (PR-AUC, F1, MCC)

Feature Engineering Techniques:

• Mathematical: Log, Box-Cox, polynomial
• Encoding: One-hot, target, frequency
• Interactions: Ratios, products
• Time: Lags, rolling stats, cyclical encoding
• Aggregations: Group statistics, rankings

Model Interpretation:

• Global: Feature importance, SHAP summary plots
• Local: SHAP values, LIME, counterfactuals
• Visualization: Comprehensive chart templates

Best Practices Built-In

✓ Proper train/test splits to avoid data leakage ✓ Cross-validation for robust evaluation ✓ Multiple metrics for comprehensive assessment ✓ Feature importance and model interpretation ✓ Documentation for reproducibility ✓ Business context integration ✓ Deployment considerations

✗ Common pitfalls actively prevented:

• Using accuracy on imbalanced data
• Test data leakage
• Ignoring outliers
• Over-engineering features
• Correlation-causation confusion

---

Project Manager Skill - Deep Dive

Core Workflows

The project-manager skill provides 5 comprehensive workflows aligned with PMBOK® standards:

1. Requirements Definition Workflow

   • Systematic requirements gathering across stakeholders
   • Functional and non-functional requirements documentation
   • MoSCoW prioritization (Must/Should/Could/Won’t Have)
   • Requirements traceability matrix
   • Change control process establishment
   • ISO/IEC/IEEE 29148 compliant templates

2. Project Plan Review Workflow

   • 10 PMBOK knowledge area checklist
   • 5 stakeholder perspective reviews
   • Gap identification and risk flagging
   • Red flag detection (e.g., no contingency, inadequate testing)
   • Comprehensive review report generation

3. Progress Reporting with EVM Workflow

   • Earned Value Management calculations (SPI, CPI, SV, CV)
   • Forecast metrics (EAC, ETC, VAC, TCPI)
   • Recovery action recommendations
   • Trend analysis and variance reporting
   • Professional progress report generation

4. Risk Management Workflow

   • Information gathering across 14 categories
   • Risk identification across 9 risk types
   • Probability × Impact scoring
   • Four response strategies (Avoid, Mitigate, Transfer, Accept)
   • Detailed response planning with preventive/detective/contingent actions
   • Risk monitoring and escalation triggers

5. Cost Estimation Workflow

   • Four estimation methods (Analogous, Parametric, Bottom-Up, Three-Point)
   • Component-based cost breakdown
   • Validation against benchmarks and historical data
   • Confidence range presentation (P10, P50, P90)
   • Contingency and management reserve allocation

Quick Start Examples

Example 1: Requirements Definition

User: "I need to create a requirements document for our new CRM system"

Claude (using project-manager skill):
1. Gathers project context through structured questions
2. Uses requirements_definition_template.md as foundation
3. Guides through functional and non-functional requirements
4. Applies MoSCoW prioritization
5. Creates traceability matrix
6. Generates comprehensive ISO-compliant requirements document

Example 2: EVM Progress Analysis

User: "Our project has these metrics: BAC=¥50M, PV=¥30M, EV=¥25M, AC=¥28M"

Claude (using project-manager skill):
1. Calculates SPI = 0.83 (17% behind schedule) 🔴
2. Calculates CPI = 0.89 (11% over budget) 🔴
3. Forecasts EAC = ¥56.2M (¥6.2M overrun)
4. Calculates TCPI = 1.14 (needs 14% efficiency improvement)
5. Status: CRITICAL (both schedule and cost issues)
6. Recommends recovery actions:
   - Fast-track activities on critical path
   - Value engineering to reduce costs
   - Immediate escalation to sponsor
7. Generates comprehensive progress report

Example 3: Risk Assessment

# Automated project health check
python scripts/project_health_check.py project_metrics.json --output report/

# Input: metrics.json with schedule, cost, quality, risk data
# Output:
# - Overall health score (0-100)
# - Status: Healthy 🟢 / At Risk 🟡 / Critical 🔴
# - Issue identification and recommendations

Example 4: Project Plan Review

User: "Review our project plan for a legacy system migration"

Claude (using project-manager skill):
1. Reviews across 10 PMBOK knowledge areas
2. Identifies gaps:
   - ❌ No knowledge transfer plan (high risk - tribal knowledge)
   - ❌ Testing only 10% of schedule (should be 25-30%)
   - ❌ No rollback plan for cutover
   - ❌ Insufficient contingency (5% vs 20% recommended)
3. Reviews from 5 stakeholder perspectives
4. Provides red flag warnings
5. Generates detailed review report with recommendations

Advanced Features

Earned Value Management (EVM):

• Complete EVM metric suite: PV, EV, AC, BAC, SV, CV, SPI, CPI, EAC, ETC, VAC, TCPI
• Three EAC calculation methods for different scenarios
• Performance index trending
• Integration with Agile (story point-based EVM)

Risk Management:

• 14 information gathering categories (scope, stakeholders, team, vendors, schedule, budget, etc.)
• 9 risk categorization types (scope, schedule, cost, quality, resource, stakeholder, technology, vendor, external)
• Quantitative risk scoring (probability × impact)
• Risk heat map visualization
• Preventive, detective, and contingent response planning
• Risk monitoring dashboard with trend indicators

Cost Estimation Methods:

• Analogous: Quick estimates based on historical projects
• Parametric: Statistical relationships (e.g., cost per story point)
• Bottom-Up: Detailed work package estimation
• Three-Point: PERT estimation with optimistic/most likely/pessimistic scenarios

Stakeholder Management:

• Stakeholder register with interest/influence matrix
• Tailored communication plans
• Engagement strategies for different stakeholder types
• Escalation paths and decision-making frameworks

PMBOK Knowledge Area Coverage

Integration Management:

• Project charter creation
• Integrated project plan development
• Change control process
• Lessons learned capture

Scope Management:

• WBS development
• Scope definition (in-scope/out-of-scope)
• Acceptance criteria definition
• Scope change management

Schedule Management:

• Activity sequencing
• Critical path analysis
• Schedule buffer allocation
• SPI tracking and recovery actions

Cost Management:

• Cost estimation (multiple methods)
• Budget baseline creation
• CPI tracking and EAC forecasting
• Contingency management

Quality Management:

• Quality standards definition
• Testing strategy (20-30% of schedule)
• Defect tracking and metrics
• Quality audits and reviews

Resource Management:

• RACI matrix creation
• Resource allocation and leveling
• Skill gap analysis
• Team development planning

Communications Management:

• Communication plan with frequency/format
• Stakeholder-specific reporting
• Escalation procedures
• Status reporting (🟢🟡🔴 indicators)

Risk Management:

• Three-phase approach (Information → Analysis → Response)
• Risk register maintenance
• Mitigation strategy development
• Risk monitoring and control

Procurement Management:

• Contract type selection (Fixed-price, T&M, Cost-plus)
• Vendor evaluation and management
• SLA definition and monitoring
• Contract administration

Stakeholder Management:

• Stakeholder identification and analysis
• Engagement strategy development
• Expectation management
• Conflict resolution

Best Practices Built-In

✓ Requirements signed off before design ✓ 15-20% schedule contingency buffer ✓ 10-20% cost contingency reserve ✓ Testing allocated 20-30% of schedule ✓ Weekly SPI/CPI tracking ✓ Risk register updated weekly ✓ C-level executive sponsor required ✓ Change control process mandatory ✓ Documentation and traceability maintained

✗ Common pitfalls actively prevented:

• Starting development without signed-off requirements
• No contingency reserves (schedule or cost)
• Inadequate testing time (<20%)
• Fixed-price contracts with unclear scope
• Ignoring critical path dependencies
• Weak executive sponsorship
• Missing change control process
• Risk management as afterthought

Real-World Examples

Example 1: Enterprise CRM Implementation (¥50M, 12 months)

• 247 requirements documented with MoSCoW prioritization
• 27 risks identified and managed (3 critical, 8 high, 16 medium/low)
• Monthly EVM tracking (SPI, CPI)
• Delivered 92% of Must Have requirements
• Finished 5% under budget
• CSAT increased from 7.2 to 8.9

Example 2: Legacy System Migration (¥95M, 20 months)

• Comprehensive risk assessment (34 risks)
• Critical mitigation: Hired retired developers for tribal knowledge
• Zero-downtime cutover with blue-green deployment
• Delivered 2 months late but within revised budget
• System performance exceeded targets (250K TPS vs 200K target)
• Stakeholder satisfaction: 9.2/10

---

Business Analyst Skill - Deep Dive

Core Workflows

The business-analyst skill provides 5 comprehensive workflows aligned with BABOK® Guide v3 standards:

1. Requirements Elicitation Workflow

   • Stakeholder identification and engagement planning
   • Elicitation technique selection (6+ techniques)
   • Structured requirements gathering sessions
   • MoSCoW prioritization (Must/Should/Could/Won’t Have)
   • Requirements documentation and review
   • Requirements validation and sign-off

2. Business Process Analysis Workflow

   • As-is process documentation with BPMN
   • Process metrics collection (cycle time, efficiency, error rate, cost)
   • Value stream mapping and waste identification (TIMWOOD)
   • Root cause analysis (5 Whys, Fishbone diagrams)
   • To-be process design with improvement opportunities
   • Process optimization recommendations

3. Stakeholder Analysis and Engagement Workflow

   • Stakeholder register creation
   • Power/Interest matrix analysis (4 quadrants)
   • Engagement strategy development per quadrant
   • RACI matrix for decision-making clarity
   • Communication plan with frequency and format

4. Business Case Development Workflow

   • Problem/opportunity definition with quantified impact
   • Options analysis with weighted scoring
   • Financial analysis (ROI, NPV, IRR, Payback Period)
   • Benefit realization planning
   • Professional business case document generation

5. Gap Analysis Workflow

   • Current state assessment (processes, systems, capabilities)
   • Future state vision definition
   • Gap identification with priority scoring
   • Roadmap development for closing gaps
   • Change impact assessment

Quick Start Examples

Example 1: Requirements Elicitation

User: "I need to gather requirements for our new employee portal"

Claude (using business-analyst skill):
1. Identifies stakeholders: HR team, employees, IT, managers
2. Creates Power/Interest matrix for engagement planning
3. Recommends elicitation techniques:
   - Workshops with HR (high interest/influence)
   - Surveys for employees (high volume)
   - 1:1 interviews with IT (technical validation)
4. Guides through structured requirements gathering:
   - User stories with acceptance criteria
   - Non-functional requirements (performance, security)
   - Business rules documentation
5. Applies MoSCoW prioritization
6. Generates comprehensive BRD using template

Example 2: Business Process Analysis

User: "Analyze our current order processing workflow - it's taking too long and has errors"

Claude (using business-analyst skill):
1. Documents as-is process with BPMN notation
   - 8 steps, 45 minutes avg processing time
   - 3 manual handoffs, 2 system switches
   - 15% error rate requiring rework

2. Identifies waste using TIMWOOD framework:
   - Waiting: 20 minutes between handoffs
   - Over-processing: Duplicate data entry in 3 systems
   - Defects: 15% error rate (data entry mistakes)

3. Performs root cause analysis:
   - Primary cause: No integration between systems
   - Secondary: Manual validation prone to errors

4. Designs to-be process:
   - Automated data integration (reduce to 2 systems)
   - Real-time validation at point of entry
   - Elimination of manual handoffs
   - Target: 10 minutes, <2% error rate

5. Quantifies improvement:
   - Time savings: 78% reduction (35 min per order)
   - Cost savings: ¥15M annually
   - Customer satisfaction: Expected +25%

6. Generates comprehensive process analysis report

Example 3: Business Case Development

# Financial analysis with sensitivity
python scripts/business_analysis.py financial \
  --investment 10000000 \
  --annual-benefit 3500000 \
  --annual-cost 500000 \
  --years 5 \
  --discount-rate 0.10 \
  --sensitivity

# Output:
# Net Annual Benefit: ¥3,000,000
# Payback Period: 3.3 years
# ROI (5 years): 50%
# NPV (5 years, 10%): ¥1,372,867
# IRR: 15.2%
#
# Sensitivity Analysis:
# Best Case (benefit +20%, cost -10%):  NPV ¥4,127,340
# Likely Case (base):                   NPV ¥1,372,867
# Worst Case (benefit -20%, cost +10%): NPV -¥1,381,606
#
# Recommendation: APPROVE - Positive NPV in likely case, acceptable risk profile

Example 4: Stakeholder Analysis

User: "Create a stakeholder analysis for our CRM implementation"

Claude (using business-analyst skill):
1. Creates stakeholder register:
   - CEO (High Power, High Interest) → Manage Closely
   - Sales Team - 50 users (High Interest, Medium Power) → Keep Informed
   - IT Ops (Medium Interest, High Power) → Keep Satisfied
   - Finance Director (Medium Interest, High Power) → Keep Satisfied

2. Analyzes attitudes:
   - CEO: Champion (actively promotes)
   - Sales Team: Neutral (wait and see)
   - IT Ops: Skeptical (concerned about support burden)
   - Finance: Supporter (sees cost benefits)

3. Develops engagement strategies:
   - CEO: Monthly 1:1 briefings, involve in key decisions
   - Sales: Bi-weekly demos, involve in UAT, address "WIIFM"
   - IT: Early technical collaboration, address concerns
   - Finance: Monthly cost reports, ROI tracking

4. Creates RACI matrix for key activities
5. Generates communication plan with frequency/format
6. Produces comprehensive stakeholder analysis document

Example 5: Data Quality Profiling

# Profile dataset for quality assessment
python scripts/business_analysis.py profile customer_data.csv --output quality_report.json

# Output:
# Data Quality Assessment
# =======================
# Overall Score: 67/100 (Needs Improvement)
#
# Completeness: 75% (25% missing values in email, phone)
# Accuracy: 60% (40% invalid email formats detected)
# Consistency: 80% (name formatting inconsistent)
# Timeliness: 50% (50% of records >2 years old)
# Validity: 85% (15% values outside expected ranges)
# Uniqueness: 90% (10% duplicate records found)
#
# Recommendations:
# 1. HIGH: Implement email validation at entry point
# 2. HIGH: Create data cleanup process for missing values
# 3. MEDIUM: Standardize name formatting
# 4. MEDIUM: Archive/purge records >2 years old

Advanced Features

BABOK® Knowledge Area Coverage:

1. Business Analysis Planning & Monitoring:

• Stakeholder analysis with engagement strategies
• Requirements management approach definition
• Change control process establishment
• Performance metrics tracking

2. Elicitation & Collaboration:

• 6+ elicitation techniques (interviews, workshops, surveys, observation, document analysis, prototyping)
• Facilitation best practices
• Conflict resolution approaches
• Collaborative decision-making frameworks

3. Requirements Life Cycle Management:

• Requirements traceability (Business Need → Business Req → Functional Req → Test Case)
• Change management and version control
• Requirements approval and sign-off processes
• Requirements reuse and templates

4. Strategy Analysis:

• SWOT analysis
• Porter’s Five Forces
• Value chain analysis
• Business model canvas
• Strategic roadmap development

5. Requirements Analysis & Design Definition:

• Requirements specification and modeling
• MoSCoW prioritization
• Gap analysis techniques
• Business rules definition
• Data flow and entity relationship modeling

6. Solution Evaluation:

• Benefits realization tracking
• Performance measurement (KPIs)
• Solution validation and acceptance criteria
• Post-implementation review

Financial Analysis Capabilities:

The business_analysis.py script provides comprehensive financial analysis:

# Calculate ROI
roi = ((total_benefit - total_cost) / total_cost) × 100

# Calculate NPV (Net Present Value)
npv = Σ(cash_flow / (1 + discount_rate)^year)

# Calculate IRR (Internal Rate of Return)
# NPV = 0, solve for discount rate

# Calculate Payback Period
payback = investment / annual_net_benefit

# Perform Sensitivity Analysis
# Best: benefit +20%, cost -10%
# Likely: base case
# Worst: benefit -20%, cost +10%

Process Analysis Techniques:

BPMN Notation:

• Swimlane diagrams with roles and handoffs
• Process flow with decision points
• System integration points
• Exception handling flows

Value Stream Mapping:

• Value-added time vs. total lead time
• Process efficiency calculation
• Waste identification (TIMWOOD framework)
• Bottleneck analysis

TIMWOOD Waste Framework:

• Transportation: Unnecessary movement of data/materials
• Inventory: Work-in-progress backlogs
• Motion: Unnecessary user actions
• Waiting: Idle time between steps
• Over-production: Creating outputs before needed
• Over-processing: Unnecessary complexity
• Defects: Errors requiring rework

Root Cause Analysis:

• 5 Whys: Iterative questioning to find root cause
• Fishbone (Ishikawa) Diagram: 6M categories (Man, Machine, Method, Material, Measurement, Mother Nature)

Data Quality Framework:

6 Dimensions of Data Quality:

1. Accuracy: Correctness of data values
2. Completeness: No missing critical values
3. Consistency: Uniformity across datasets
4. Timeliness: Data currency and relevance
5. Validity: Values within expected ranges
6. Uniqueness: No unintended duplicates

Requirements Engineering Best Practices

Effective Requirements Characteristics (SMART):

• Specific: Clear and unambiguous
• Measurable: Success criteria defined
• Achievable: Technically and organizationally feasible
• Relevant: Aligned with business objectives
• Time-bound: Delivery expectations clear

Requirements Format:

FR-001: Automated Order Validation
Priority: Must Have
User Story: As a sales rep, I want the system to automatically validate
           customer credit limits so that I can process orders faster and
           avoid bad debt

Acceptance Criteria:
- Given an order with total > customer credit limit
- When the order is submitted
- Then the system displays credit limit warning within 2 seconds
- And requires manager approval before proceeding

Business Value: Reduce bad debt by 80% (from ¥5M to ¥1M annually)
Dependencies: BR-003 (Credit check API integration)
Source: VP Sales, CFO

MoSCoW Prioritization:

• Must Have: Critical for go-live, non-negotiable
• Should Have: Important but not critical, defer if needed
• Could Have: Desirable, include if capacity allows
• Won’t Have: Out of scope for this release

Requirements Traceability Matrix:

| Business Need | Business Req | Functional Req | Design Element | Test Case | Status |
|---------------|-------------|----------------|----------------|-----------|--------|
| Reduce bad debt | BR-001 | FR-005, FR-006 | Credit API | TC-025 | Approved |

Stakeholder Management Best Practices

Power/Interest Matrix:

           High Interest
                │
                │
 Keep Informed  │   Manage Closely
  (Bi-weekly    │   (Weekly 1:1,
   updates)     │    Key decisions)
                │
Low ────────────┼──────────── High
Power           │            Power
                │
    Monitor     │   Keep Satisfied
  (Monthly      │   (Monthly reports,
   summaries)   │    Avoid surprises)
                │
           Low Interest

RACI Matrix Example:

| Activity | Executive | Business Owner | BA | Project Manager | Dev Team |
|----------|-----------|----------------|-----|-----------------|----------|
| Approve scope | A | C | I | R | I |
| Gather requirements | I | C | R | C | I |
| Review design | I | C | A | C | R |
| UAT | I | R | C | A | I |
| Go-live decision | A | R | C | C | I |

Legend: R=Responsible, A=Accountable, C=Consulted, I=Informed

Communication Plan Considerations:

• Executive Sponsor: Strategic alignment, ROI, decisions needed (Monthly)
• End Users: “What’s in it for me?”, training, support (Bi-weekly)
• IT Teams: Technical approach, integration, support model (Weekly)
• Finance: Cost tracking, benefit realization (Monthly)

Business Case Components

Problem/Opportunity Statement:

• Quantified current state impact
• Qualitative impacts (morale, reputation)
• Cost of inaction

Options Analysis:

| Option | Weighted Score | Investment | NPV (5yr) | Risk |
|--------|---------------|-----------|----------|------|
| Option 1: COTS | 85/100 | ¥10M | ¥5.2M | Medium |
| Option 2: Custom | 78/100 | ¥15M | ¥6.8M | High |
| Option 3: Hybrid | 92/100 | ¥12M | ¥7.1M | Low |

Benefit Realization Plan:

• Quantitative benefits with measurement approach
• Qualitative benefits
• Benefit timeline (quick wins vs. long-term)
• Baseline and target metrics

Risk Assessment:

• Probability × Impact scoring
• Mitigation strategies
• Contingency reserves

Best Practices Built-In

✓ Stakeholder identification early and comprehensive ✓ Requirements elicited from multiple perspectives ✓ MoSCoW prioritization for scope management ✓ Requirements traceability maintained ✓ Financial analysis with sensitivity testing ✓ Process analysis with quantified improvements ✓ Data quality assessment before analysis ✓ Documentation following BABOK standards ✓ Change management approach integrated

✗ Common pitfalls actively prevented:

• Starting solution design before understanding problem
• Missing key stakeholders (discovering late)
• Assuming requirements without validation
• No requirements prioritization (everything is critical)
• Business cases without financial rigor
• Process analysis without data/metrics
• Ignoring change management and adoption
• Poor requirements documentation and traceability

Real-World Examples

Example 1: Order Processing Automation (¥8M investment)

• Problem: 45 min/order, 15% error rate, ¥25M annual cost
• Solution: Automated workflow with system integration
• Results:
  • Processing time: 10 min (78% reduction)
  • Error rate: <2% (87% reduction)
  • Annual savings: ¥15M
  • Payback: 6.4 months
  • ROI: 187% (3 years)
  • Customer CSAT: 7.1 → 8.8

Example 2: CRM Implementation (¥12M investment)

• 127 requirements documented (MoSCoW: 45/52/30/0)
• 23 stakeholders identified, engagement strategies per quadrant
• Financial analysis: NPV ¥18.2M (5 years), IRR 24.1%
• Process improvements: Sales cycle reduced 30%
• Delivered 94% of Must Have + Should Have requirements
• User adoption: 89% within 60 days
• Benefit realization: 112% of projected (over-achieved)

Example 3: Data Quality Initiative

• Problem: 40% of customer data incomplete/inaccurate
• Data profiling revealed: Completeness 60%, Accuracy 55%
• Root causes: No validation at entry, multiple legacy systems
• Solution: Data quality rules, stewardship program, consolidation
• Results:
  • Data quality score: 60 → 92 (over 12 months)
  • Marketing campaign effectiveness: +45%
  • Duplicate account reduction: 85%
  • Annual savings: ¥3.2M

🏭 Production Schedule Optimizer

File: skill-packages/production-schedule-optimizer.skill

Optimizes weekly production schedules for manufacturing facilities (central kitchens, food factories, production lines).

When to use:

• Creating weekly production schedules for central kitchens or food factories
• Auto-calculating production frequency based on shelf life
• Optimizing room capacity and staff allocation
• Estimating staff requirements and designing shift plans
• Identifying bottlenecks and improving schedule balance

Key Features:

• Greedy Bin-Packing scheduling with deterministic sort/tie-break
• 4-CSV input system (products, demand, rooms, staff)
• Shelf-life based production frequency calculation
• Staff requirement estimation with buffer coefficient
• PSO-E/W alert system for validation and quality monitoring
• Markdown timetable output

Scripts:

• generate_schedule.py - Weekly schedule generation CLI
• estimate_staff.py - Staff requirement estimation CLI

📋 Shift Planner

Generates weekly employee shift schedules from roster and requirements using constraint-satisfaction greedy assignment.

When to use:

• Creating individual shift schedules from staff requirements
• Auto-assigning employees based on qualifications, availability, and labor constraints
• Verifying shift coverage at 30-minute granularity
• Analyzing fairness across employees (hours deviation, weekend distribution)
• Working with production-schedule-optimizer output for downstream shift planning

Key Features:

• Constraint-satisfaction greedy assignment (difficulty-first slot ordering)
• 3-CSV input system (roster, requirements, optional shift patterns)
• Hard constraints: max hours/days, qualifications, consecutive days, rest hours
• Soft constraints: avoid days, preferred patterns, weekend balance
• 30-minute coverage verification with break exclusion
• Fairness metrics: hours deviation, weekend distribution, avoid violations
• SFT-E/W alert system (8 error codes, 10 warning codes)
• Built-in 5 shift patterns (FULL_8H, EARLY_8H, LATE_8H, SHORT_6H, HALF_4H)
• 25 test cases

Scripts:

• generate_shifts.py - Shift generation CLI

---

🔍 Incident RCA Specialist

Systematically analyzes incidents to identify root causes and develop corrective action plans. Focuses on organizational process improvement, not log-level debugging (use log-debugger for that).

When to use:

• Conducting post-incident reviews
• Creating incident reports and RCA documentation
• Developing corrective action plans with SMART criteria
• Performing organizational root cause analysis (5 Whys, Fishbone, FTA)
• Building incident timelines with TTD/TTR/TTM/TTRe metrics

Key Features:

• 8 workflows: Information Gathering → Timeline → Impact Assessment → 5 Whys → Fishbone → FTA → Corrective Actions → Report
• 5 Whys with branching technique and human error decomposition (never stop at “operator error”)
• Fault Tree Analysis with AND/OR gates, minimal cut sets, SPOF identification
• 3D Prevention Framework (Detect/Defend/Degrade)
• P0-P4 severity matrix with SLA violation evaluation
• Mermaid gantt-based timeline with TTD/TTR/TTM/TTRe calculation
• Bilingual report templates (Japanese/English)
• Corrective action tracker with SMART criteria validation

---

🔄 Iterative Design Assistant

File: skill-packages/iterative-design-assistant.skill

Tracks design iteration history for documents and presentations, understands context from previous change requests, and applies consistent styling decisions across multiple revision cycles.

When to use:

• User references a previous design decision (“前回も色で良いんだけど”, “like last time”)
• Multiple revision cycles on the same document/presentation
• Need to track which design elements were changed and why
• Applying consistent styling across related documents
• Reviewing design history to understand document evolution
• User asks to “undo” or “revert” to a previous design state

Key Features:

• Session-local design decision log with JSON schema
• 5 decision categories: color, typography, layout, content, style
• Contextual reference resolution (Japanese/English patterns)
• Design token extraction and management
• Bidirectional traceability (decisions ↔ elements)
• Markdown/JSON history report generation

Scripts:

• design_log.py - CLI for init, record, query, search, apply, history, token, resolve

References:

• design-decision-methodology.md - Best practices for tracking and applying design decisions

Example Use Cases:

• “Use the same blue as before” → Resolves to most recent color decision
• “前回と同じフォントで” → Applies previous typography decision
• “Generate a history of all design changes” → Markdown report with timeline

---

📝 Technical Spec Writer

Creates structured technical specifications bridging requirements and implementation. Generates screen designs, API specs, DB designs, sequence diagrams, and state transition diagrams with Mermaid.

When to use:

• Creating functional specifications from BRD/requirements
• Designing screen layouts with UI element and event tables
• Creating REST API design documents
• Designing database schemas with ER diagrams
• Generating sequence and state transition diagrams

Key Features:

• 7 workflows: Requirements Intake → Screen Design → API Design → DB Design → Sequence Diagram → State Transition → Document Assembly
• IEEE 830 / ISO 29148 compliant specifications
• ID numbering system (SCR-xxx, API-xxx, TBL-xxx, SEQ-xxx, STS-xxx)
• Mermaid diagram patterns (sequence, state, ER, flowchart)
• REST API design guide with error codes, pagination, authentication
• DB design guide with normalization, indexing, audit columns
• Traceability matrix (REQ → SCR/API/TBL mapping)
• Bilingual templates (Japanese/English)

---

📋 Operations Manual Creator

Creates structured operations manuals and SOPs using the STEP format (Specific/Target/Expected/Proceed) with ANSI Z535-inspired caution/warning classification.

When to use:

• Creating operations manuals for business systems
• Writing standard operating procedures (SOPs)
• Creating user guides with step-by-step instructions
• Building troubleshooting guides with escalation procedures

Key Features:

• 6 workflows: Scope Definition → Operations Inventory → Procedure Writing → Caution/Warning Notes → Troubleshooting → Assembly
• STEP format: Specific action, Target UI element, Expected result, Proceed confirmation
• ANSI Z535 classification: DANGER/WARNING/CAUTION/NOTE with placement rules
• Operation ID system (OP-xxx) with dependency mapping
• Symptom-Cause-Resolution troubleshooting tables
• L1/L2/L3 escalation procedure templates
• Screenshot placeholders and annotation guidance
• Bilingual templates (Japanese/English)

---

🎯 CX Error Analyzer

Analyzes error/exception scenarios from a customer experience perspective, scoring them on 6 axes and prioritizing improvements using an Impact vs Effort matrix.

When to use:

• Evaluating error scenarios from CX perspective
• Improving error message quality and recovery flows
• Prioritizing error UX improvements
• Creating CX-focused error analysis reports

Key Features:

• 5 workflows: Error Inventory → Multi-Axis Evaluation → CX Scoring → Priority Matrix → Report
• 6-axis evaluation: Impact Severity (25%), Frequency (20%), Recovery Ease (15%), Message Quality (15%), Emotional Impact (10%), Business Cost (15%)
• CX Score tiers: Critical (4.0-5.0) / Significant (3.0-3.9) / Moderate (2.0-2.9) / Minor (1.0-1.9)
• Impact vs Effort matrix with Quick Wins identification
• Error UX best practices (message design, recovery flows, emotional design)
• CX metrics reference (CES/CSAT correlation, support cost, churn risk)
• ROI calculation for error UX improvements
• Error classification taxonomy and user journey mapping

---

📊 Management Accounting Navigator

Routes management accounting inquiries to the appropriate analysis skill by auto-classifying user requests into 12 management accounting domains. COSO/IMA framework aligned.

When to use:

• Determining which management accounting analysis to perform
• Routing budget variance, CVP, cost accounting inquiries
• Navigating across management accounting domains

Key Features:

• 12-domain auto-classification with routing to specialized skills
• COSO/IMA management accounting framework alignment
• Bilingual domain classification templates (Japanese/English)
• Covers: Budget variance, CVP, standard costing, KPI design, ABC, make-or-buy

---

💰 MA: Budget-Actual Variance Analysis

Analyzes budget vs actual variances with automatic favorable/unfavorable classification by account type (revenue/expense), variance decomposition, and root cause hypothesis generation.

When to use:

• Monthly/quarterly budget vs actual comparison
• Variance decomposition (price variance, quantity variance)
• Management reporting with variance ranking

Key Features:

• Auto favorable/unfavorable determination by account type
• Price and quantity variance decomposition
• Significance ranking and root cause hypothesis
• CSV data upload support
• Bilingual report templates (Japanese/English)

---

📈 MA: CVP / Break-Even Analysis

Performs Cost-Volume-Profit analysis including break-even point calculation, contribution margin analysis, margin of safety evaluation, and target profit simulation. Supports multi-product analysis.

When to use:

• Calculating break-even point (sales amount/volume)
• New business or pricing change profitability simulation
• Fixed cost reduction / variable cost improvement impact analysis
• What-if scenario analysis

Key Features:

• Fixed/variable cost structure analysis
• Break-even point calculation (amount and volume)
• Contribution margin ratio and margin of safety
• Target profit required sales simulation
• Multi-product CVP analysis support
• Bilingual templates (Japanese/English)

---

🏭 MA: Standard Cost Variance Analysis

Analyzes variances between standard (planned) costs and actual costs, decomposing into price and quantity variances across material, labor, and overhead categories.

When to use:

• Manufacturing cost variance analysis
• Standard costing system operation and reporting
• Cost reduction activity effectiveness measurement

Key Features:

• Price and quantity variance decomposition
• Material, labor, and manufacturing overhead categorization
• Favorable/unfavorable auto-determination
• Responsible department identification
• Root cause hypothesis generation
• CSV data upload support
• Bilingual templates (Japanese/English)

---

🎯 Presentation Reviewer

Reviews presentation materials from the audience perspective, evaluating content clarity, visual design, logical flow, engagement factors, and Marp technical compatibility.

When to use:

• Reviewing draft presentations before delivery
• Getting objective feedback on slide quality
• Checking Marp template compatibility
• Improving presentation effectiveness

Key Features:

• 5 evaluation axes: Content clarity, visual design, logical flow, engagement, technical compatibility
• Audience perspective review methodology
• Actionable improvement recommendations

---

🔧 MARP Layout Debugger

File: skills/marp-layout-debugger/

Diagnoses and fixes common MARP slide layout issues including whitespace problems, box alignment, bullet formatting inconsistencies, and CSS rendering issues. Provides visual diff comparisons and automated fixes.

When to use:

• MARP slides have unexpected whitespace or spacing issues
• Box elements are misaligned or overlap incorrectly
• Bullet points have inconsistent indentation or formatting
• Content overflows slide boundaries
• CSS styles render differently than expected
• Need to validate MARP CSS against best practices

Key Features:

• 5 issue categories: Whitespace (WS), Alignment (AL), Bullets (BL), Overflow (OF), CSS (CS)
• 16 specific issue types with severity classification and auto-fix capability
• Automated fix application with safety-first approach (auto-fix vs manual review)
• Visual diff report generation showing before/after comparison
• Non-destructive analysis with backup support

---

wbs-review-assistant v1.0 (2026-03-19)

• WBS Excel file review against requirements documents and hearing sheets
• Automatic gap detection with requirement traceability analysis
• Excel cell annotation with severity-based conditional formatting (Critical/Major/Minor)
• Structural validation: WBS code hierarchy, phase organization, milestone checks
• Content quality checks: effort estimates, task descriptions, acceptance criteria
• Hearing notes cross-check for decision alignment
• Three output formats: Annotated Excel, Markdown summary, JSON gap analysis
• Readiness score calculation (0-100) with Go/No-Go recommendation
• 14 common WBS issue patterns with auto-detection
• Bilingual support (Japanese/English WBS and requirements)
• Marp-specific technical checks
• Best practices checklist reference

---

Skill Idea Miner

File: skill-packages/skill-idea-miner.skill

Mine Claude Code session logs for skill idea candidates, score them for novelty, feasibility, and work utility, and maintain a prioritized backlog for downstream skill generation.

When to use:

• Weekly automated pipeline run for skill idea generation
• Manual backlog refresh from recent coding sessions
• Dry-run to preview candidates without LLM scoring

Key Features:

• Session log extraction from ~/.claude/projects/ JSONL files
• LLM-based scoring (novelty, feasibility, work utility, composite)
• Backlog management with YAML persistence and dedup
• mine_session_logs.py - Extract candidates from session logs
• score_ideas.py - Score and merge candidates into backlog

---

Skill Designer

File: skill-packages/skill-designer.skill

Design new Claude skills from structured idea specifications. Generates comprehensive Claude CLI prompts that create complete skill directories following repository conventions.

When to use:

• The skill auto-generation pipeline selects an idea from the backlog
• Bootstrapping a new business/professional skill from a JSON idea specification
• Quality review of generated skills requires awareness of the scoring rubric

Key Features:

• Design prompt generation from JSON idea specs
• Repository convention compliance (SKILL.md frontmatter, directory structure)
• Integration with dual-axis-skill-reviewer scoring rubric
• build_design_prompt.py - Generate design prompt from idea JSON

---

Codebase Onboarding Generator

File: skill-packages/codebase-onboarding-generator.skill

Automatically analyze a codebase and generate comprehensive CLAUDE.md documentation for future Claude Code sessions. Identifies project type, common commands, build processes, test patterns, directory structure conventions, and architectural decisions.

When to use:

• Setting up Claude Code for a new project that lacks CLAUDE.md
• Generating initial project documentation for AI assistants
• Refreshing outdated CLAUDE.md files after significant project changes
• Creating standardized onboarding documentation for team codebases
• Analyzing unfamiliar codebases to understand structure and conventions

Key Features:

• Automatic project type detection (Python, Node.js, Rust, Go, Java, etc.)
• Command extraction from package.json, pyproject.toml, Makefile, Cargo.toml
• Framework detection (FastAPI, Django, React, Vue, Spring, etc.)
• Environment variable detection from .env files and source code
• Directory structure analysis with semantic descriptions
• analyze_codebase.py - Main analysis script with JSON and CLAUDE.md output

Supported Project Types:

• Python (pyproject.toml, setup.py, requirements.txt)
• Node.js (package.json with npm/yarn/pnpm/bun detection)
• Rust (Cargo.toml)
• Go (go.mod)
• Java (Maven pom.xml, Gradle build.gradle)
• Ruby (Gemfile)
• PHP (composer.json)
• .NET (*.csproj, *.sln)

---

🌍 Timezone-Aware Event Tracker

File: skill-packages/timezone-aware-event-tracker.skill

Track and correlate events across multiple timezones with automatic conversion. Maintains awareness of regional time differences (PST/CST/EST/JST and others), handles daylight saving time (DST) transitions, and generates time-normalized reports for distributed team incident analysis.

When to use:

• Analyzing incidents or logs from distributed systems spanning multiple timezones
• Correlating events from teams in different regions (US West, US East, Japan, etc.)
• Creating unified timelines from events recorded in different local times
• Investigating issues where timestamp confusion led to coordination failures
• Scheduling or reviewing cross-regional meetings and handoffs

Key Features:

• Automatic timezone detection from timestamp suffixes (PST, EST, JST, etc.)
• DST transition handling with ambiguity detection and warnings
• Event correlation within configurable time windows
• Multi-timezone timeline reports (Markdown and JSON)
• Pattern classification (cascading failure, rapid sequence, simultaneous)
• timezone_event_tracker.py - CLI for parse, correlate, report, and dst-check commands

Supported Timestamp Formats:

• ISO 8601 with offset (2024-03-15T10:30:00-07:00)
• Datetime with abbreviation (2024-03-15 10:30:00 PST)
• Common log format (15/Mar/2024:10:30:00 -0700)
• IANA timezone names (America/Los_Angeles, Asia/Tokyo)

---

📱 QR Code Generator

File: skill-packages/qr-code-generator.skill

Generate QR code images from arbitrary strings (URLs, text, contact information, etc.) using Python’s qrcode library. Supports customization of size, margin (quiet zone), foreground/background colors, and error correction levels. Batch generation mode allows creating multiple QR codes from a CSV or JSON input file.

When to use:

• Generate a QR code from a URL, text, or data
• Create multiple QR codes in batch from a file
• Create QR codes with specific customization (size, colors, error correction)
• Encode contact information (vCard) as a QR code

Key Features:

• Single QR code generation with full customization
• Batch processing from CSV or JSON files
• vCard generation for contact information
• Configurable error correction levels (L, M, Q, H)
• Custom colors (named colors, hex, RGB)
• Automatic preview with Claude’s image display
• generate_qr.py - CLI for single and batch QR code generation

Parameters:

• Box size (module pixel size): 5-20+ pixels
• Border (quiet zone): minimum 4 modules
• Error correction: L (7%), M (15%), Q (25%), H (30%)
• Fill/background colors: named, hex (#RRGGBB), or RGB format

---

🚪 Completion Quality Gate Designer

File: skill-packages/completion-quality-gate-designer.skill

Design quality gates, exit criteria, evidence requirements, and exception governance for each project phase. Separates “Implemented”, “Verified”, “Accepted”, “Released”, and “Exception-approved” states to prevent premature completion claims.

When to use:

• Defining Definition of Done per project phase
• Designing release readiness gates with evidence requirements
• Standardizing verification commands across CI, developers, and reports
• Establishing exception governance for incomplete items
• Reconciling test metrics between reports and dashboards

Key Features:

• 7-phase gate design workflow (scope → vocabulary → exit criteria → evidence → commands → exceptions → expression control)
• 5 templates: Quality Gate Matrix, Definition of Done, Exception Register, Release Readiness, Evidence Ownership Matrix
• Completion vocabulary separation framework (Implemented/Verified/Accepted/Released/Exception-approved)
• Metrics reconciliation guide with single source of truth principles

---

🔍 Hidden Contract Investigator

File: skill-packages/hidden-contract-investigator.skill

Extract implicit contracts from existing code, functions, and modules before reuse. Identifies mismatches between expected and actual behavior — return types, side effects, environment dependencies, and scope shadowing.

When to use:

• Reusing legacy functions in new features
• Investigating functions where names or comments don’t match behavior
• Extracting return type, side effect, and exception contracts from code
• Assessing reuse risk before implementation
• Designing contract verification tests

Key Features:

• 6-step investigation workflow (target → surface contract → actual contract → mismatch classification → reuse judgment → verification design)
• 6-category mismatch taxonomy (Naming, Type, Scope, State, Environment, Hidden Side Effect)
• 5-level reuse risk classification (A: as-is through E: do not reuse)
• Contract test idea generation with pytest examples

---

🛡️ Safe By Default Architect

File: skill-packages/safe-by-default-architect.skill

Convert recurring dangerous implementation patterns into safe architectural defaults and enforceable standards. Defines forbidden patterns, approved replacements, common layers, static rule candidates, and exception governance.

When to use:

• Same dangerous code patterns keep recurring across controllers/pages
• Establishing deny-by-default authorization, ORM-only queries, service-layer I/O
• Creating static analysis rules (lint, semgrep) from defect patterns
• Documenting forbidden-to-safe pattern mappings with code examples
• Designing Architecture Decision Records for safety standards

Key Features:

• 6-step workflow (pattern aggregation → danger classification → safe standard definition → default decision → common layer design → operational deployment)
• 7-category safe pattern catalog (query, auth, file, persistence, datetime, dependency, idempotency)
• Forbidden-to-safe mapping with before/after code examples
• Static rule candidate templates with false positive assessment

---

🔗 Cross Module Consistency Auditor

File: skill-packages/cross-module-consistency-auditor.skill

Map change impact across all affected modules, flows, reports, and APIs. Audit consistency rules for aggregation totals, sign inversion, status transitions, and copy-paste propagation.

When to use:

• One specification change affects multiple screens, reports, or APIs
• Deploying the same logic to multiple flows (e.g., 6 POS transaction types)
• Verifying refund/void/correction flows maintain sign consistency
• Reviewing copy-paste implementations efficiently (canonical + diff strategy)
• Ensuring report totals match drill-down details

Key Features:

• 6-step workflow (change kernel → impact lens → impact map → consistency rules → copy propagation strategy → test checklist)
• 7-category consistency rule catalog (aggregation, status transitions, sign inversion, tax/rounding, visibility, naming, report alignment)
• Copy propagation review strategy with canonical-plus-diff approach
• Cross-module test checklist generation

---

🧪 Production Parity Test Designer

File: skill-packages/production-parity-test-designer.skill

Design test hierarchies that catch production-specific failures before deployment. Allocates failure modes to appropriate test tiers, eliminates proxy metrics, and creates adversarial regression backlogs.

When to use:

• PR CI is too light to catch production-specific failures
• SQLite/PostgreSQL or other DB dialect gaps exist
• UI shows success but database persistence is unverified
• Mocks hide runtime import errors or dependency gaps
• Defining what belongs in unit vs integration vs smoke vs E2E

Key Features:

• 7-step workflow (gap inventory → failure modes → tier allocation → proxy elimination → smoke suite → adversarial backlog → command map)
• Production gap catalog covering DB dialect, dependencies, env vars, timezone, OS, mock vs real, serialization
• Adversarial test patterns (injection, bypass, traversal, fake success, import mismatch, etc.)
• Standard command map separating local/CI/staging/release test commands

---

🔒 Web Server Security Reviewer

File: skills/web-server-security-reviewer/

Phase 1 web server (nginx/apache, Linux-centric) security configuration review. Validates target with target_profile.yaml, verifies evidence integrity via MANIFEST.txt + manifest_attestation.txt, then performs read-only inspection across 9 axes either over SSH or against a pre-collected manifest.

When to use:

• Pre-deployment / pre-go-live web server hardening review
• Periodic security posture audit of nginx or apache hosts
• Reviewing pre-collected evidence (MANIFEST + attestation) without direct SSH access
• Documenting findings with severity (Exploitability × Blast Radius × Service Criticality) and observation_status

Key Components:

• references/checklist_9axes.yaml — 9-axis review checklist (OS/resources/logs/network/services/auth/monitoring/backup/certs)
• references/command_reference.yaml — read-only command catalog with 6-tier guardrails
• references/finding_taxonomy.md / severity_criteria.md — severity & finding model
• references/observation_status.md — managed unverified-finding states
• references/role_extensions/ — role-specific extensions (edge / internal / public-facing + schema)
• assets/target_profile_template.yaml / manifest_template.txt / manifest_attestation_template.txt — input templates
• assets/report_template.md / action_plan_template.md / self_review_template.md — output templates
• assets/wizard_answers_example.json — sample wizard answers for build_target_profile.py
• references/interview_wizard.md — 7-stage declarative spec for the Phase 0 interactive wizard
• scripts/build_target_profile.py — wizard answers JSON → target_profile.yaml builder (defaults / derivations / hard-fail gates)
• scripts/verify_skill.py — internal integrity verifier (22 checks, includes wizard pipeline smoke)
• scripts/verification_run.py — review run orchestrator

Key Features:

• 6-tier command guardrails (allowed / conditional / conditional_sensitive / ask-first / exceptional_sensitive_read / forbidden) prevent destructive or PII-exposing commands
• 3-axis severity scoring with explicit observation_status separation for “not verifiable from evidence”
• evidence_dir / raw_evidence_store split keeps secrets out of the report directory
• MANIFEST integrity (SHA256, attestation truthfulness levels) detects tampered evidence
• Windows / IIS hard-fail in v1 (out-of-scope) so you don’t get a misleading partial review
• Mandatory self-review template before finalizing findings
• 21 forbidden commands enforced; SSH redirect handling for sensitive paths
• Cross-references all template / command IDs (verifier confirms zero unresolved refs)

---

🧠 Fable Thinking

File: skills/fable-thinking/

Fableクラスの思考プロセスを任意のモデルで再現する思考スキャフォールド（knowledge-only）。「視野を広げてから絞る」「自分の結論を一度攻撃してから出す」という思考の順序を、8フェーズと通過ゲートで強制する。モデルの規模に依存しない手続きなので、Opus や Sonnet でも深く幅広い検討を経た結論を再現できる。

When to use:

• 設計判断、技術選定、移行判断など、間違えたときの手戻りコストが高いタスク
• 原因が自明でない調査、トラブルシュート、データ分析
• 提案書・戦略文書・レビューなど、結論の説得力が成果物の価値を決めるタスク
• 「深く考えて」「幅広い視野で」「多角的に」「Fableのように」と指示されたとき

Key Components:

• SKILL.md — 8フェーズの思考プロトコル本体（トリアージ → 問いの再構築 → 前提の棚卸し → 視野の拡張 → 深掘り → 反証 → 収束 → セルフレビュー）
• references/thinking-protocol.md — 各フェーズの質問バンクと、浅い回答と深い回答を比較する実例
• references/lenses.md — 視野を強制的に広げる12のレンズ集（時間軸、二次効果、逆転、プレモーテム、ベースレート、可逆性ほか）とタスク種類別の推奨セット
• references/anti-patterns.md — 10の失敗パターンと対策の早見表（アンカリング、メニュー回答、ヘッジ逃げ、記憶による断定ほか）

Key Features:

• 各フェーズに Gate（通過条件）を設け、弱いモデルが素通りしがちな検討ステップを強制する
• 「質的に異なる3案が揃うまで収束禁止」でアンカリングを構造的に防止
• 反証フェーズで結論を仮固定して攻撃し、負けたら選択肢展開に戻るループ設計
• 出力契約: 結論先頭・推奨1つ・トレードオフ・反証条件の4点セット。思考の足場は出力に見せない
• /fable-thinking で直接呼び出し、またはエージェント定義から Follow the fable-thinking protocol で参照可能

---

🧒 ELI5

File: skills/eli5/

トピックを「大きな絵と最小限の言葉」だけで説明する、単一のHTMLアーティファクトを生成するスキル。名前は5歳児向けだが、狙いは子ども向けの語彙で話すことではなく、読み手が持っていない前提知識を取り除くこと。想定読者は「そのテーマを一度も聞いたことがない賢い大人」で、平易さと正確さを両立させる。

When to use:

• 「ELI5」「簡単に説明して」「図で説明して」「初心者向けに」「素人にもわかるように」と依頼されたとき
• 技術的な内容を、非技術者の上長・顧客・家族に説明する必要があるとき
• ビジュアル説明資料、図解ウォークスルー、初心者が追える1枚ものを求められたとき
• 一度説明したが伝わらず、もっと平易な版を求められたとき

Key Components:

• SKILL.md — 説明設計の方針と、生成するHTMLアーティファクトの構成規定

Key Features:

• 冒頭に1文サマリ、数枚の大パネル、締めの「だから何なのか」1行という固定構成
• CSSとSVGをすべてインライン化した自己完結型HTML。外部スクリプト・スタイル・画像を読み込まない
• 正確な技術図より、うまい例えを絵にしたものを優先する
• 全体を1分で読み切れる分量に制限。段落を書きそうになったら図に置き換える
• 出力言語は依頼者に追従。指定された聞き手が別言語を読む場合はそちらを優先
• 対象外は厳密な技術仕様書と、既に分野を理解している聴衆向けのプレゼン資料

Credits: MITライセンスの eli5 プラグイン（作者 Thariq Shihipar）をもとに拡張。

---

🔥 Grill Me（詰めるインタビュー）

File: skills/grill-me/

要件・企画・計画・提案・アイデアを、1問ずつの容赦ないインタビューで詰めきり、最後にブリーフとして書き起こすスキル。依頼者の頭の中にある構想は本人が思っているよりぼやけている、という前提に立ち、そのぼやけを質問で洗い出す。対象は文書ではなく会話なので、事前に何かが書かれている必要はない。

When to use:

• 「grill me」「grill this」「clarify」「詰めて」「詰めたい」「壁打ち」「要件を固めたい」「企画を練りたい」「計画を整理したい」「アイデアをぶつけたい」「突っ込んで」「質問して」「何が決まってないか教えて」と言われたとき
• 動き出す前に、自分の考えを叩いておきたいとき
• 計画やアイデアを漠然と説明して、次に何をすべきか尋ねられたとき
• 仕様書・提案書・資料をこれから書くが、材料が固まっていないとき

Key Components:

• SKILL.md — インタビューの進め方。設計ツリーとフロンティアの考え方、質問の作り方、ぼやけの検出パターン、終わらせ方
• references/lenses.md — 対象の種類ごとの質問の切り口（要件／企画／計画／アイデア／提案の5種）
• references/output-format.md — ブリーフのテンプレート
• NOTICE — 上流の著作権表示とMIT許諾文

Key Features:

• 対象を設計ツリーとして捉え、前提が確定した判断（フロンティア）のうち最も多くを解きほぐす1問だけを聞く
• 1ターンに1問。 番号付きの質問を並べると流し読みされるが、選択肢付きの1問なら3秒で実際の判断が返る
• すべての質問に推奨回答を添える。 推奨のない質問は宿題、推奨のある質問はその場で受け入れられる判断になる
• 曖昧な数量表現、曖昧なスコープ、未定義の成功、登場人物・制約の欠落、順序の未記述、借り物の確信の7パターンからぼやけを検出
• 調べられる事実はスキルが調べ、依頼者にしか出せない事実だけを聞く
• 8〜10問でチェックポイントを置き、続けるかここでまとめるかを依頼者に選ばせる
• 成果物は、詰めきった要約・用語集・判断ログ・未決事項・次の一手からなるブリーフ1枚

Credits: MITライセンスの grilling / domain-modeling スキル（作者 Matt Pocock）と、本リポジトリの commands/clarify.md をもとに構成。

---

Roadmap

Future skills planned for this library:

• [x] Business Analyst - Requirements analysis, process mapping, business case development (COMPLETED)
• [ ] NLP Specialist - Text analysis, sentiment, topic modeling
• [ ] Computer Vision Analyst - Image classification, object detection
• [ ] Deep Learning Engineer - Neural networks, transfer learning
• [ ] AB Testing Statistician - Experimental design, statistical testing
• [ ] Business Intelligence Analyst - Dashboard creation, KPI tracking
• [ ] Data Engineer - ETL pipelines, data quality monitoring
• [ ] Salesforce Consultant - CRM configuration, workflow automation, requirement gathering

Version History

jev-artifact-style-review: レポートのエスケープと入力サイズ上限 (2026-10-23)

• review.md: 原文引用をコードブロックに、文書名・位置・保護語句をコード表記に入れる。原文中の Markdown や HTML が見出し・コード範囲・HTML要素を新しく作らない
• --context のファイルは 64,000 バイト、compare.py に渡す review.json は 8,000,000 バイトを超えると、解析する前に拒否する

task-runner: 実行中タスクの誤解放を防止、dry-run と計画承認の判定順を明確化 (2026-10-23)

• doing/ にタスクがある間は、開始から60分を超えても次のタスクを始めない。開始日時の欠落・不正・未来時刻も停止の疑いとして表示し、人が前回の終了を確認してから移す手順に変更
• 手順4の判定順を dry-run → 計画承認 → 作業に固定。dry-run では計画の確認が必要なタスクも blocked/ に送らず todo/ に戻す。mode・外部操作・計画確認・計画承認の組み合わせ表を追加

codex-reviewer: GPT-6.1-Sol へ更新 (2026-10-23)

• 全レビュータイプと両プロファイルの既定モデルを gpt-5.5 から gpt-6.1-sol に変更。推論レベルは従来どおり xhigh（quick-review は medium）
• analyze_review.py: リストマーカー付きの指摘（- **重要度**: ...）も解析できるようにした

japanese-clear-writing v1.1.0 / ste-writer v1.0 / task-runner v1.0 (2026-10-04)

• japanese-clear-writing: 日本語文書の新規執筆・書き直し・添削・評価の4モードと統合モード。6項目ルーブリックと scripts/score.py
• ste-writer: ASD-STE100 の主要14ルールで英文を執筆・書き換え。full / soft の2段階
• task-runner: フォルダ型タスクチケットの定期実行と、scripts/build_dashboard.py による看板ボード生成
• 配布済みの版を取り込み。pre-commit による行末空白・末尾改行の修正と、Python 2ファイルの ruff format を適用
• japanese-clear-writing の frontmatter から author を削除
• task-runner: 存在しない日付をファイル名に持つチケットでダッシュボード生成全体が落ちる不具合を修正し、日付なしとして扱うようにした。--now が不正な場合は分かるエラーで終了する。同日内の処理順がファイル名の昇順であることと、## 結果 を毎回置き換えることをドキュメントに明記
• score.py と build_dashboard.py にテストを追加し、CI の test-skills に登録

jev-artifact-style-review v1.0 (2026-09-20)

• 日本語中心・英語対応の文体レビュー。「AIっぽさ」を執筆者の推定ではなく、修正可能な編集上の欠点として扱う
• 文体8軸（決まり文句、同義反復、過剰な構造化、文型・語尾の単調さ、口調・敬語の不一致、直訳調、誇張・装飾、対話AI定型の残存）と内容品質4軸を独立に採点
• 0〜100の文体違和感指数は文体8軸のみを集計。AI執筆確率でも読者の違和感率でもない
• 採点と根拠選択は Jev、妥当性判断・説明・修正文案はホストのエージェントという役割分担
• 用途別 profile 7種類とルーブリック外部化（assets/rubric.json）
• scripts/review.py（抽出・dry run・採点）と scripts/compare.py（同一条件での再評価比較）
• dry run の出力は extracted.json / context.json / plan.json / request_preview.json。採点する実行ではこれに review.md / review.json / writer_handoff.json / run_status.json が加わる
• reviewer_feedback.md は CLI ではなくホストのエージェントが生成する。既存ディレクトリへの上書きなし、POSIX権限600/700
• 校正状況は未検証。同梱テスト64件は合成応答で、実APIには接続していない
• 配布元 v1.0.0 を取り込み。本リポジトリのCIに合わせて Python 11ファイルを整形した。ロジックは変更しておらず、整形後もテスト64件の合格を再確認
• TEST_REPORT.md のヘッダー3行は行末2スペース改行を箇条書きへ変換。pre-commit が行末スペースを除去し3行が1段落に結合されるため。文言は不変
• requirements.txt の pypdf の下限を >=5 から >=6.10 へ引き上げ。配布元の範囲は CVE-2026-40260 の影響版を許容するため。Python 3.10 と pypdf 6.19.0 でテスト64件の合格を確認
• MANIFEST.sha256 は上記3点を反映して再生成したため、配布元 zip のハッシュとは一致しない

grill-me v1.0 (2026-09-08)

• Relentless one-question-at-a-time interview that sharpens requirements, business plans, project plans, proposals, and raw ideas
• Models the subject as a design tree; asks only the single frontier question that unblocks the most of it, then recomputes
• Every question carries a recommended answer, which turns a question from homework into a decision the user can accept instantly
• Seven fuzz-detection patterns: vague quantifiers, ambiguous scope, undefined success, missing actors, missing constraints, unstated sequencing, borrowed certainty
• Facts are looked up rather than asked; only facts about the user’s own world (budget, headcount, what the boss said) are put to them
• Checkpoint after ~8-10 questions so the user decides how deep to go
• Output is a single brief: sharpened summary, glossary, decision log, and open items
• references/lenses.md carries question angles for 5 subject types; references/output-format.md carries the brief template
• Adapted from the MIT-licensed grilling / domain-modeling skills by Matt Pocock and this repository’s commands/clarify.md

vendor-support-ticket-tracker v1.0 (2026-09-01)

• Track vendor support tickets and RMA cases across multiple vendors from a local YAML database
• 8-state ticket lifecycle: open → acknowledged → in-progress → awaiting-parts / awaiting-customer → escalated → resolved → closed
• SLA targets by priority (critical 4h, high 24h, medium 72h, low 5 days) with business-hours calculation and breach warnings
• Stale-ticket detection with a configurable threshold, plus escalation history tracking
• scripts/ticket_manager.py CLI: init / create / update / report / stale / escalate / show / list
• references/ticket-lifecycle.md documents state transitions, SLA guidelines, and escalation procedures
• Markdown status reports identifying tickets that need follow-up

eli5 v1.0 (2026-08-30)

• Plain-language visual explainer that renders a topic as a single self-contained HTML artifact
• Big pictures, very few words: one-sentence opener, a few large panels, one concrete example, a closing “so what” line
• Inline CSS and SVG only — no external scripts, stylesheets, or images
• Targets “a smart adult who has never heard of this” rather than a literal five-year-old, keeping the explanation both simple and accurate
• Bilingual trigger coverage (ELI5 / explain like I’m five / 簡単に説明して / 図で説明して / 初心者向けに / 素人にもわかるように); output follows the user’s language, or the named audience’s
• Adapted from the MIT-licensed eli5 plugin by Thariq Shihipar

fable-thinking v1.0 (2026-07-07)

• Knowledge-only thinking scaffold that reproduces Fable-class deep reasoning on any model (Opus / Sonnet)
• 8-phase protocol with exit gates: Triage → Question Reframing → Assumption Audit → Option Expansion → Deep Dive → Adversarial Self-Critique → Calibrated Convergence → Self-Review
• 12 perspective lenses (time horizon, second-order effects, inversion, pre-mortem, base rate, reversibility, etc.) with task-type recommendation sets
• Anti-pattern catalog: 10 shallow-reasoning failure modes with countermeasures (anchoring, menu answers, hedging, literal-answer trap, etc.)
• Worked example contrasting shallow vs scaffolded answers (SQLite→PostgreSQL migration case)
• Output contract: conclusion-first, single recommendation + trade-offs + falsification conditions; scaffold stays in thinking
• Invocable via /fable-thinking, natural-language triggers (深く考えて / 多角的に / Fableのように), or agent definitions

email-thread-summarizer v1.0 (2026-06-14)

• Email thread analysis skill for extracting status, action items, timeline, and decisions
• Multi-format support: Gmail JSON, Outlook/Graph API, EML, MBOX
• Participant role detection (initiator, responder, CC)
• Action item extraction with deadline detection using NLP patterns
• Decision identification and tracking
• Superseded information detection (corrections, updates, contradictions)
• 5-state thread status classification (active, resolved, stale, escalated, awaiting_response)
• Staleness detection for cached summaries with recommendation engine
• Markdown summary generation with timeline, action items, and decisions
• 69 unit tests covering all core functionality

web-server-security-reviewer v1.1 (2026-05-02)

• Phase 0 interactive interview wizard for building target_profile.yaml via AskUserQuestion
• New references/interview_wizard.md — 7-stage declarative spec (stage_id / field / prompt / header / choices / required / condition / normalize_to)
• New scripts/build_target_profile.py — wizard answers JSON → target_profile.yaml builder
  • Hard-fail gates (windows / iis), machine_id_sha256 strict validation, ISO8601 enforcement
  • Auto-derivations: delete_after, source_confidence, role_extensions from preset roles
  • Conditional requirements: ssh.* when ssh_direct, incident_id/summary when active_incident=true
  • CSV normalization for array fields (out_of_scope / access_approvers / rotation_dependencies)
  • normalize_optional_null() for ssh.bastion (collapses “” / “none” / “null” / “なし” / “n/a” → None)
  • target.custom_role substitution when target.role == "custom"
• New assets/wizard_answers_example.json smoke fixture
• 78 unit tests (16 new vs v1.0 baseline)
• verify_skill.py expanded to 22 checks (added section H: wizard pipeline smoke)

web-server-security-reviewer v1.0 (2026-05-01)

• Phase 1 web server (nginx/apache, Linux-centric) security configuration review skill
• 9-axis checklist: OS / Resources / Logs / Network / Services / Authentication / Monitoring / Backup / Certificates
• 6-tier command guardrails: allowed / conditional / conditional_sensitive / ask-first / exceptional_sensitive_read / forbidden
• 3-axis severity model: Exploitability + Blast Radius + Service Criticality
• Two evidence-collection modes: direct SSH (read-only) and pre-collected manifest (MANIFEST.txt + manifest_attestation.txt with SHA256 integrity)
• observation_status taxonomy keeps unverified findings managed separately
• evidence_dir / raw_evidence_store separation; mandatory self-review before finalization
• Windows/IIS hard-fails in v1 (out-of-scope); rhel/nginx is the primary supported profile
• 8 templates (target_profile, manifest_attestation, finding_table, report, action_plan, evidence_directory_layout, self_review, manifest)
• 12 reference docs + 4 role_extension YAMLs (generic_edge / generic_internal / generic_public_facing + _schema)
• verify_skill.py integrity script (17/17 PASS at v1.0; 22/22 PASS at v1.1) and verification_run.py runner

meeting-minutes-writer v1.0 (2026-04-30)

• Generate meeting minutes from transcripts/notes with built-in self-review loop (max 3 iterations)
• 5 Mandatory Checks per iteration: Internal Contradictions, Consistency, Action-Item Omissions, Speaker-Name Errors, Date/Day-of-Week Errors
• MANDATORY date verification via python3 -c "import datetime; ..." — never memory-based
• Severity model (HIGH/MEDIUM/LOW); HIGH findings blocking completion
• Completion report surfaces remaining HIGH findings and * To be confirmed items after iteration 3
• Complements meeting-minutes-reviewer (review-only) and video2minutes (transcribe→write)
• Resources: output_format.md, self_review_checklist.md, minutes_template_en.md, minutes_template_ja.md, findings_report_template.md (bilingual)

internal-email-composer v1.0 (2026-04-17)

• Compose professional internal emails for coordination tasks
• 6 supported scenarios: vendor RFQ, task delegation, status update, follow-up, escalation, info request
• Bilingual support (Japanese/English) with culturally-adapted content
• Business etiquette compliance (敬語/keigo for Japanese, professional tone for English)
• 3 urgency levels (normal, high, urgent) with appropriate subject prefixes
• CLI script compose_email.py with JSON and Markdown output formats
• Template engine with variable substitution for key points, deadlines, attachments
• Comprehensive reference guides for email templates and business etiquette

iterative-design-assistant v1.0 (2026-03-29)

• Session-local design decision log with JSON schema (schema v1.0)
• 5 decision categories: color, typography, layout, content, style
• CLI commands: init, record, query, search, apply, history, token, resolve
• Contextual reference resolution for Japanese and English patterns
• Design token extraction and management with category namespacing
• Bidirectional traceability (decisions ↔ elements)
• Markdown and JSON history report generation

vendor-procurement-coordinator v1.0 (2026-04-18)

• End-to-end vendor procurement workflow coordination
• Orchestrates vendor-rfq-creator and vendor-estimate-creator skills
• Project initialization with standard directory structure (rfq/, quotes/, estimates/, communications/)
• Vendor management (add, edit, remove, CSV import) with status tracking
• Quote response logging with amount, currency, delivery date, validity period
• Email template system (Japanese/English RFQ emails, reminder templates)
• Vendor comparison report generation with price scoring and analysis
• Procurement status tracking (project and vendor lifecycle states)
• Timeline event logging for complete audit trail
• Python CLI scripts: init_procurement.py, manage_vendors.py, track_responses.py, compare_quotes.py
• YAML-based procurement configuration (procurement.yaml)

ai-bpo-proposal-generator v1.0 (2026-04-19)

• Generate AI-powered BPO service proposals for Japanese companies in the US market
• Service module selection based on industry and pain points (15 service modules, 3 industry bundles)
• ROI calculation with current/future state analysis, NPV, payback period
• Phased implementation roadmap generation with milestones
• Bilingual (Japanese/English) proposal document generation
• Service categories: Finance & Accounting, HR & Payroll, Customer Support, Data Processing, Procurement
• Client intake questionnaire template with 8 sections
• Industry benchmarks for automation rates, error reduction, and cost savings

inbox-triage-summarizer v1.0 (2026-06-13)

• Periodic inbox scan for project/client-based email categorization
• 4-way action classification: FYI, requires-response, requires-action, blocked-waiting
• Project matching via sender domain, subject keywords, and thread labels
• Thread correlation and staleness tracking (warning/follow-up/escalation thresholds)
• Urgency scoring with deadline detection (URGENT, ASAP, EOD, explicit dates)
• Export format compatible with email-triage-responder for response drafting
• Scan history tracking for delta-based processing
• YAML-based project mapping configuration
• JSON and Markdown summary report outputs

email-triage-responder v1.0 (2026-04-21)

• Email inbox triage and response draft generation skill
• Urgency detection: time-sensitive keywords, deadlines, escalation markers
• Importance scoring: VIP sender patterns, recipient position, content signals
• Topic classification: 8 categories (client, vendor, internal, meeting, FYI, approval, escalation, delegation)
• Eisenhower Matrix quadrant assignment (Q1-Q4 prioritization)
• Contextual draft response generation with tone/language adaptation
• Response status tracking: pending, draft_ready, sent, delegated, archived
• JSON and Markdown report output formats
• Integration with gogcli for Gmail and MCP for Outlook

meeting-minutes-reviewer v1.0 (2026-03-26)

• Review meeting minutes for completeness, action item clarity, and decision documentation
• 5-dimension quality scoring: Completeness (25%), Action Items (25%), Decisions (20%), Consistency (15%), Clarity (15%)
• Action item validation: owner, deadline, description completeness checks
• Decision documentation validation: context, rationale, alternatives considered
• Source material consistency verification (hearing sheets, agendas)
• Vague language detection with specific suggestions
• JSON and Markdown report output formats
• CLI with review_minutes.py script

action-status-updater v1.0 (2026-04-04)

• Natural language action item tracking for Japanese and English
• Intent detection (completed, delegated, deferred, in-progress) with regex patterns
• Person, channel, and keyword extraction from status updates
• Persistent YAML state with full history tracking
• CLI tool with init, add, update, report, export, and list commands
• Integration guide for daily-comms-ops workflow
• 52 tests covering NL parsing and state management

purchase-request-generator v1.0 (2026-03-30)

• Generate formal IT/hardware purchase request documents from informal specifications
• Cost-benefit analysis with ROI, NPV, and payback period calculations
• Weighted vendor comparison matrices with scoring and ranking
• MARP prese