A Comprehensive Classification Matrix for the OpenClaw Skill Library: 227 Skills × Four Categories × a Full Business-Process Map
Core proposition: When the skill library swells to 227, how do you make every skill findable and usable without trimming? The answer is to build a two-dimensional classification matrix organized by business category × business stage. Data: 227 skills · 4 major categories · 10 business stages · 720 cross-references Methodology: full skill scan → description keyword classification → two-dimensional cross-mapping → matrix visualization
Introduction: Why a Skill Library Needs Matrix Management
OpenClaw's skill system uses a mechanism of "keyword matching on the description → automatic context injection". As the number of skills grew from 50 to 227, a fundamental problem emerged:
When a user says "help me build a website", how does the system precisely match
design-taste-frontend+ui-ux-pro-max+Frontend Designout of 227 skills, rather thanfirecrawl-website-design-cloneorvideo-frames?
The answer requires cross-location across two dimensions:
- Functional category (what the skill "does")
- Business stage (when the skill is "used")
The matrix established in this article is precisely the infrastructure that solves this problem.
1. The Four Functional Categories
Based on the actual business needs of Junze Zhiku, the 227 skills are divided into four major categories:
| Category | Skill Reference Count | Core Positioning |
|---|---|---|
| 🛠️ Everyday use / core system tools | 348 | Essential for daily operation: search, communication, memory, system management |
| 💰 Finance-related | 66 | Hong Kong stock research, M&A, valuation, trading |
| 💻 Coding-related | 238 | Coding, design, engineering, testing, deployment |
| 📦 Other | 68 | Unclassified or low-frequency use |
⚠️ Skills can be classified into multiple categories. For example,
deploy-vercelbelongs both to "everyday core tools" and to "coding-related".
2. The Ten Business Stages
Skills within each category are arranged by business-process stage:
📋 Planning and Analysis → 🔍 Research and Discovery → 🎨 Design and Prototyping → 💻 Development and Implementation
↓
🧪 Testing and Diagnostics → 📊 Analysis and Reporting → 🚀 Deployment and Release
↓
🧠 Memory and Storage → 💬 Communication and Management → ⚙️ System Operations
3. 🛠️ Everyday Use / Core System Tools
Essential for daily operation: memory, communication, system management, search | 348 references | 11 stages
🔍 Search and Research (76 skills)
Internet search, data collection, intelligence gathering:
firecrawl-search firecrawl-scrape firecrawl-parse firecrawl-crawl firecrawl-map firecrawl-agent tavily-search cross-validated-search deep-research x-research web-learner web_search web_fetch
💬 Communication and Management (51 skills)
The Feishu ecosystem, instant messaging, email notifications:
feishu-im feishu-calendar feishu-task feishu-doc feishu-sheet feishu-bitable feishu-wiki feishu-chat email-report email-monitor discord whatsapp handoff weekly-update
⚙️ System Operations (34 skills)
Configuration management, containerization, monitoring, security:
gateway cron healthcheck docker-compose docker-essentials doctor backup cloudflare-tunnel agent-access-control agent-rate-limiter agent-analytics agent-audit
🧠 Memory and Storage (12 skills)
Long-term memory, knowledge management, log organization:
memory-hub obsidian openclaw-auto-dream vector-memory agentmemory memory-recall memory-consolidate memory-checkpoint
🚀 Deployment and Release (11 skills)
deploy-vercel cc-projects-list clawhub github
4. 💰 Finance-Related
Financial analysis, Hong Kong stock research, M&A, valuation, trading | 66 references | 10 stages
🔍 Search and Research (15 skills), the Core HK Stock DD Pipeline
ak-hk-stock-dd ak-bdd ak-company-dd sdd hk-stock-dd ak-sdd-list financial-analyst firecrawl-market-research
📊 Analysis and Reporting (11 skills), Financial Analysis and Report Generation
ak-financial-analyst dd-business-report financial-overview dcf-valuation ipo junze-valuation
📋 Planning and Analysis (10 skills), Deal Strategy and Project Management
junze-deal-sourcing buyer-list deal-screening deal-tracker deal-sourcing catalyst-calendar dd-checklist dd-meeting-prep
🎨 Design and Prototyping (8 skills), Deal Documents
cim-builder teaser investment-proposal white-knight process-letter strip-profile ic-memo returns-analysis
5. 💻 Coding-Related
Coding, design, engineering, testing, deployment | 238 references | 10 stages
📋 Planning and Analysis (26 skills), the Core of Engineering Discipline
grill-with-docs grill-me improve-codebase-architecture zoom-out to-prd to-issues triage design-taste-frontend software-architect prd-to-ddd-design
🎨 Design and Prototyping (46 skills), the Frontend Design Ecosystem
design-taste-frontend ui-ux-pro-max frontend-design Frontend Design html-ppt html-ppt-course-module dashboard blog-post editorial-* landing-pages email-marketing web-prototype-* animejs gsap css-animations canvas
💻 Development and Implementation (40 skills), the Coding Toolchain
tdd claude-code-bridge claude-code deepseek-bridge codewhale-bridge tailwind vibe-coding setup-pre-commit git-guardrails-claude-code migrate-to-shoehorn scaffold-exercises node-connect node-inspect-debugger
🧪 Testing and Diagnostics (35 skills), Quality Assurance
diagnose tdd python-debugpy setup-pre-commit firecrawl-qa review
🚀 Deployment and Release (14 skills)
deploy-vercel cc-projects-list github clawhub
6. 📦 Other
68 references | 11 stages
Includes QwenCloud AI services (qwencloud-text qwencloud-image-generation qwencloud-video-generation qwencloud-audio-tts), Understand-Anything code analysis (8 understand-* skills), general tools (1password nano-pdf trello openhue blucli), and more.
7. Skill Router v2 Routing Strategy
Based on the matrix above, Skill Router's routing rules can be precise down to category × stage:
| User Request Keywords | Matched Category | Matched Stage | Recommended Skills |
|---|---|---|---|
| Look up stocks, research, DD | 💰 Finance | 🔍 Search | ak-hk-stock-dd + tavily-search |
| Valuation, DCF, financials | 💰 Finance | 📊 Analysis | ak-financial-analyst + dcf-valuation |
| Build a website, design, UI | 💻 Coding | 🎨 Design | design-taste-frontend + ui-ux-pro-max |
| Write features, develop | 💻 Coding | 💻 Development | tdd + grill-with-docs |
| Debug, error | 💻 Coding | 🧪 Diagnostics | diagnose |
| Deploy, ship | 🛠️ Everyday | 🚀 Deployment | deploy-vercel |
| Memory, organize, review | 🛠️ Everyday | 🧠 Memory | memory-hub + openclaw-auto-dream |
| Feishu, notification, email | 🛠️ Everyday | 💬 Communication | feishu-im + email-report |
8. Skill Health Insights
| Insight | Data | Recommendation |
|---|---|---|
🔴 The firecrawl-* family is bloated | 15 → only 3 used | Trim down to search/scrape/parse |
🔴 lark-* Feishu tools are idle | 15 → only 3-4 used | Keep the high-frequency ones, enable the rest on demand |
🔴 qwencloud-* is entirely idle | 9 → API Key not activated | Remove temporarily, restore after activation |
🟡 The hyperframes-* animation family | 20 → used only once | Keep 3 core ones |
| 🟢 The finance pipeline is complete | 36 → full coverage from DD → analysis → report | Keep |
| 🟢 Engineering discipline is in place | grill + tdd + diagnose + architecture | Promote wider use |
9. Summary
227 skills are not the problem; the problem is having no map. This classification matrix solves three core pain points:
- Finding skills: know the task type → consult the matrix → find the corresponding skill combination
- Preventing overlap: skills of the same kind are clear at a glance, avoiding redundant installations
- Routing optimization: provides Skill Router with precise matching rules
The next step is to integrate this matrix into Skill Router v3, achieving "automatic detection of task category + business stage → precise skill combination recommendation".
This article is based on the results of an OpenClaw skills check scan. Skill count: 227 (as of 2026-06-04) Classification method: description keywords + directory names, dual matching
More in Evidence
- A Reality Check on Decision Models: Why They Seem Miraculous Online but We Measured Only 54%: A Full Comparison of JEV / LAYA / KEV / CLM-8B and a Deployment Formula
- The "Non-Text-Generating Model": Jev and the New System One Category, and How Agent Architecture Changes When AI Only Answers Multiple Choice
- WeChat Open Source WeMM-Embedding Deep Dive: The Multimodal Embedding Model Topping MMEB-v2, Can It Run on Your Mac?
- A Source-Level Architectural Dissection of DeepSeek Harness: How an Everything-Is-a-Plugin Agent Framework Is Built