268 AI coding assistant skills, organized across 12 workflow layers. Sources include Anthropic official, FRM, SKC, LRN, SKA, and other mainstream AI coding frameworks.
Each skill is annotated with ⭐1–5, indicating general applicability in daily AI-assisted development:
| ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐ | ⭐ |
|---|---|---|---|---|
| Nearly everyone | Most developers | Common in specific scenarios | Specialized needs | Highly niche |
Work Session
│
├─ 00-foundation · Foundation Layer ──── Always-on infrastructure (16)
├─ 01-discover · Discovery Layer ──── Pre-work: research, brainstorm, perspective (20)
├─ 02-define · Definition Layer ──── Pre-code: specs, plans, architecture (17)
├─ 03-build · Build Layer ──── Implementation (12 subdirs by language/domain)
│ ├─ /universal Universal methodology (22)
│ ├─ /python Python / ML (7)
│ ├─ /js-ts JavaScript / TypeScript (14)
│ ├─ /jvm Kotlin / Java / Spring / Quarkus (12)
│ ├─ /php PHP / Laravel (3)
│ ├─ /mobile Swift / Flutter / Android (10)
│ ├─ /dotnet C# / F# / .NET (3)
│ ├─ /systems Go / Rust / C++ / Perl (9)
│ ├─ /data-infra Databases / Docker / K8s (10)
│ ├─ /ui-motion Animation / Accessibility / UI Engineering (6)
│ ├─ /agents AI Agent Engineering (21)
│ ├─ /agents AI Agent Engineering (17)
│ └─ /orchestration Multi-Agent Orchestration Templates (1)
├─ 04-assure · Quality Layer ──── Pre-commit: review, testing, security (38)
├─ 05-release · Release Layer ──── Go-live: CI, releases, deployment (5)
├─ 06-reflect · Reflect Layer ──── Post-release: retros, maintenance, docs (5)
├─ 07-share · Share Layer ──── Docs, video, writing, social (22)
├─ 08-extend · Extend Layer ──── MCP, automation, third-party integrations (31)
├─ 09-design · Design Layer ──── UI/UX design exclusive (17)
└─ 11-vertical · Verticals (30): healthcare, finance, networking, business ops, career
Total: 268 skills (optimized from 351 → 322 for quality, −29; restructured via PROMPTS.md three-mode refactor, −13; further reduced to 309, then −13)
This round's optimizations:
- Removed
10-LRN(−15; meaningless to non-LRN users) - Merged
cavemanintoLZY(−1; LZY is a better-benchmarked superset) - Merged the Token trio →
token-management(−2) - Merged the Motion quartet →
motion-design(−3) - Merged 4 framework quality trios →
{fw}-quality(−8: django/springboot/quarkus/laravel) - Restructured
03-buildinto 12 language/domain subdirectories (discoverability: from "132 flat files" → "clearly grouped")
This round's refactoring (PROMPTS.md three modes):
- Pattern A: 9 language testing skills (python/kotlin/rust/cpp/golang/react/csharp/fsharp/perl-testing) → 1
universal-testing(−8) - Pattern B:
java-coding-standards+cpp-coding-standards→coding-standards(−1, moved to 00-foundation) - Pattern C: 5
orch-*orchestration skills → 1master-orchestrator(−4) - Total: −13 files; added
REFACTOR-SUMMARY.mddocumenting decision rationale
Purpose: Always-active baseline infrastructure for every session — behavioral guidelines, guardrails, context management, token control.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
using-skills |
⭐⭐⭐⭐⭐ | harness | Master entry point for this skill collection; establishes a skill-first workflow with a 12-layer quick-reference table |
ultimate-guidelines |
⭐⭐⭐⭐⭐ | Karpathy+harness | Ultimate coding guidelines: Karpathy anti-pattern checklist × simplicity-first × surgical edits × harness engineering principles |
coding-standards |
⭐⭐⭐⭐⭐ | SKC (refactored) | Externalized coding standards: read project-root .clauderules / .cursorrules / decisions.md first, follow strictly, override built-in defaults. Falls back to minimal baseline (naming, immutability, single responsibility) when no project rules exist. Replaces java-coding-standards and cpp-coding-standards |
developer-guardrails |
⭐⭐⭐⭐ | mattpocock+LRN+SKC (3-in-1) | Three-layer safety: Git hook intercepts dangerous commands + pre-commit quality gate + production system protection (enhanced SKC safety-guard) |
session-state |
⭐⭐⭐⭐ | harness+LRN (3-in-1) | Session state management: /context-save to checkpoint, /context-restore to resume, /session to initialize and load decisions.md |
ck |
⭐⭐⭐⭐ | SKC | Project-level persistent memory: auto-loads project context each session, hook-driven, zero manual effort |
context-engineering |
⭐⭐⭐⭐⭐ | agent-skills | Context engineering: context quality is the largest lever on AI output quality — 5-layer context hierarchy (rules → specs → source → errors → history), Context Packing strategy, MCP integration, confusion management, anti-pattern list |
knowledge-capture |
⭐⭐⭐ | LRN+SKC (merged) | Extract architectural patterns, anti-patterns, and tech debt from codebases; store in memory/learnings.jsonl; includes SKC knowledge-ops multi-storage-layer management |
strategic-compact |
⭐⭐⭐ | SKC | Smart context compression advice: auto-prompts manual compaction at logical task breakpoints to prevent context overflow |
context-budget |
⭐⭐⭐ | SKC | Context window consumption audit: analyze usage by agent, skills, MCP servers, and rule files; provide optimization recommendations |
token-budget-advisor |
⭐⭐⭐ | SKC | Token usage advisor: track session token consumption, forecast costs, suggest savings strategies |
cost-tracking |
⭐⭐⭐ | SKC | Read Claude Code token usage, spend, and budget reports from a local cost-tracker database |
continuous-learning-v2 |
⭐⭐ | SKC | Intuition-based learning system: observe sessions via hooks, extract atomic lessons and auto-store them |
handoff |
⭐⭐⭐ | mattpocock | Compress session into a handoff document for seamless pickup by the next agent or session |
caveman |
⭐⭐⭐ | mattpocock | Extreme compression communication mode: omit pleasantries and redundant explanations, reduce token usage by ~75% |
LZY |
⭐⭐⭐⭐ | LZY | Lazy senior engineer mode: YAGNI minimalism — stdlib-first, no unrequested abstractions, shortest working diff, 3 intensity levels (lite/full/ultra). Complements caveman (LZY manages code, caveman manages prose) |
LZY-audit |
⭐⭐⭐ | LZY | Audit codebase for "over-engineering" debt: identify unnecessary abstractions, unused configs, single-implementation interfaces |
LZY-debt |
⭐⭐ | LZY | Manage LZY debt: track where minimal principles were violated and plan remediation |
LZY-help |
⭐⭐ | LZY | LZY usage guide: intensity switching, boundary explanations, how to combine with caveman |
Purpose: Divergent thinking and information gathering before formal work begins — research, brainstorming, questioning assumptions, broadening perspective.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
ideation |
⭐⭐⭐⭐ | skills+openclaw+mattpocock (3-in-1) | Three-mode creative engine: Brainstorm (free divergence) + YC Office Hours (relentless assumption-challenging / energetic collaboration) + Grill-Me (probe until consensus) |
deep-research |
⭐⭐⭐⭐ | SKC | Multi-source deep research: firecrawl + exa MCP parallel search across web, codebase, papers; synthesize structured reports |
exa-search |
⭐⭐⭐⭐ | SKC | Neural semantic search: search web, code, company info via Exa MCP; more semantically precise than keyword search |
search-first |
⭐⭐⭐⭐ | SKC | Search before coding: find existing tools, libraries, patterns; prevent reinventing the wheel; applicable to all development scenarios |
documentation-lookup |
⭐⭐⭐⭐ | SKC | Fetch latest official docs via Context7 MCP; avoid using stale API info from training data |
codebase-onboarding |
⭐⭐⭐⭐ | SKC | Analyze unfamiliar codebases; generate structured onboarding guides with architecture diagrams, data flows, and key modules |
product-lens |
⭐⭐⭐ | SKC | Product diagnosis: validate the "why" before building; run product diagnostics; stress-test solutions from a user perspective |
zoom-out |
⭐⭐⭐ | mattpocock | Step back for the big picture; identify what's really being built; invoke when lost in details |
grill-with-docs |
⭐⭐⭐ | mattpocock | Challenge the current approach using official documentation; sharpen terminology, surface incorrect assumptions |
teach |
⭐⭐⭐ | mattpocock | Deep explanation mode: break down complex concepts into progressive teaching within the workspace context |
interview-me |
⭐⭐⭐⭐ | agent-skills | Requirements elicitation interview: surface real user needs through structured probing before writing any plan or code — clarify ambiguities, surface assumptions, confirm success criteria |
idea-refine |
⭐⭐⭐⭐ | agent-skills | Idea refinement: three-phase creative flow (diverge → converge → output one-pager) — HMW problem statements, 5–8 variant exploration, hidden assumption excavation, "Not Doing" list. Complements ideation (ideation = group divergence, idea-refine = single-idea deepening) |
council |
⭐⭐⭐ | SKC | Four-voice advisory panel: convene four perspectives (optimist/pessimist/pragmatist/innovator) for ambiguous decisions, trade-offs, and go/no-go calls |
research-ops |
⭐⭐⭐ | SKC | Evidence-first current-state research workflow: gather the latest factual information before acting |
market-research |
⭐⭐⭐ | SKC | Market research: competitive analysis, investor due diligence, industry intelligence; includes research frameworks and data analysis |
competitive-platform-analysis |
⭐⭐ | SKC | In-depth competitive platform analysis: feature comparison, pricing strategy, UX evaluation |
competitive-report-structure |
⭐⭐ | SKC | Competitive report structure template: standardized framework and writing guide for competitive analysis reports |
brand-discovery |
⭐⭐ | SKC | Brand discovery workflow: distill brand positioning and core value proposition through user interviews and market analysis |
scientific-thinking-literature-review |
⭐⭐ | SKC | Systematic literature review: literature search and evaluation workflow for academic, biomedical, technical, and scientific domains |
scientific-thinking-scholar-evaluation |
⭐⭐ | SKC | Academic work evaluation: structured review framework for papers, proposals, literature reviews, and methodologies |
Purpose: Transform vague ideas into actionable specifications — PRDs, ADRs, architecture design, plans, investment materials.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
planning |
⭐⭐⭐⭐⭐ | harness+skills | Plans.md generation and management + universal implementation planning methodology; upstream of executing-plans |
to-prd |
⭐⭐⭐⭐⭐ | mattpocock+LRN | Convert conversation context into a PRD and publish to an Issue Tracker; includes 5-phase Spec methodology |
to-issues |
⭐⭐⭐⭐ | mattpocock | Break down plans or PRDs into independent, claimable GitHub Issues |
plan-review |
⭐⭐⭐⭐ | openclaw+LRN (4-dimension collection) | Four-dimensional plan review: Engineering + CEO (11 sections) + Design + DevEx; make plans airtight before implementation |
intent-driven-development |
⭐⭐⭐⭐ | SKC | Intent-driven development: transform vague product requirements into scoped, verifiable acceptance criteria |
api-design |
⭐⭐⭐⭐ | SKC | REST API design standards: resource naming, status codes, pagination, filtering, error formats, versioning strategy |
architecture-decision-records |
⭐⭐⭐⭐ | SKC | Auto-capture architectural decisions during Claude Code sessions; generate structured ADR files |
hexagonal-architecture |
⭐⭐⭐ | SKC | Hexagonal architecture (Ports & Adapters) design, implementation, and refactoring; emphasizes domain boundaries and dependency inversion |
blueprint |
⭐⭐⭐ | SKC | System blueprint generation: from requirements description to complete technical blueprint including component diagrams, interface specs, and data models |
product-capability |
⭐⭐⭐ | SKC | Transform PRD intent and roadmap requirements into implementation-ready capability decomposition specs |
plan-orchestrate |
⭐⭐⭐ | SKC | Read plan documents, decompose steps, design agent chains for each step; suited for SKC multi-agent workflows |
benchmark-methodology |
⭐⭐⭐ | SKC | Benchmarking methodology: design repeatable performance measurement frameworks; prevent micro-benchmark misinterpretation |
planning-with-files |
⭐⭐⭐⭐⭐ | planning-with-files | File-persistent planning (Manus-style): write task_plan.md/findings.md/progress.md to disk as working memory; auto-restore after /clear and context loss via Claude Code hooks (UserPromptSubmit/PreToolUse/PostToolUse/Stop/PreCompact). Supports parallel tasks, autonomous mode, gated mode, SHA-256 plan proofs (tamper-proof) |
pm-brain |
⭐⭐⭐ | pm-brain | PM second brain: interview-style initialization, writes 57 template files, establishes hypothesis/decision/stakeholder/roadmap systems |
investor-materials |
⭐⭐ | SKC | Create and update fundraising materials: pitch decks, one-pagers, investment memos, accelerator applications |
investor-outreach |
⭐⭐ | SKC | Write investor cold emails, intro summaries, follow-up emails, update emails, and other investor communication materials |
marketing-campaign |
⭐⭐ | SKC | End-to-end marketing campaign planning: audience research, positioning strategy, channel selection, content planning, performance tracking |
Purpose: The actual coding phase — 121 skills organized into 12 subdirectories by language/domain. Quick navigation: go directly to the subdirectory matching your tech stack.
| Skill | ⭐ | Description |
|---|---|---|
test-driven-development |
⭐⭐⭐⭐⭐ | Complete TDD collection (skills + mattpocock + SKC + agent-skills, 4-in-1) |
systematic-debugging |
⭐⭐⭐⭐⭐ | Systematic debugging (skills + mattpocock + openclaw, 3-in-1) |
executing-plans |
⭐⭐⭐⭐ | Execute Plans.md: from serial single-task to full-team parallel (harness + skills) |
parallel-agents |
⭐⭐⭐⭐ | Parallel sub-agent scheduling with two-phase review |
incremental-implementation |
⭐⭐⭐⭐⭐ | Vertical-slice incremental development; thinnest end-to-end slice each iteration |
source-driven-development |
⭐⭐⭐⭐ | Source-document-driven (Context7 MCP); prevents API hallucinations |
doubt-driven-development |
⭐⭐⭐⭐ | Launch adversarial fresh-context review for every decision |
prototype |
⭐⭐⭐⭐ | Rapid prototyping; speed over perfection |
auth |
⭐⭐⭐⭐ | Auth + payment scaffolding (Clerk/Supabase/Stripe) |
crud |
⭐⭐⭐⭐ | Auto-generate CRUD interfaces |
git-workflow |
⭐⭐⭐⭐ | Git workflow standards (SKC + agent-skills) |
using-git-worktrees |
⭐⭐⭐ | Git Worktree multi-task parallelism |
error-handling |
⭐⭐⭐⭐ | Cross-language error handling patterns |
deprecation-and-migration |
⭐⭐⭐ | Phased deprecation and user migration |
benchmark-optimization-loop |
⭐⭐⭐ | Recursive performance optimization loop |
api-connector-builder |
⭐⭐⭐ | Build new API connectors by matching existing patterns |
FRM-loop |
⭐⭐ | Long-running task loop; ScheduleWakeup for resumption |
content-hash-cache-pattern |
⭐⭐⭐ | SHA-256 content-hash caching of expensive computations |
regex-vs-llm-structured-text |
⭐⭐⭐ | Decision framework: regex vs LLM for parsing structured text |
code-tour |
⭐⭐ | Generate CodeTour step-by-step code walkthrough files |
inherit-legacy-style |
⭐⭐ | Analyze and preserve legacy codebase style consistency |
universal-testing |
⭐⭐⭐⭐⭐ | Universal testing meta-skill (9-in-1): inject runtime parameters (runtime/test_runner/mock_library/coverage_tool) to activate language-specific idioms while preserving RED→GREEN→REFACTOR philosophy, assertion patterns, and test layers (Unit/Integration/Contract/E2E). Replaces python/kotlin/rust/cpp/golang/react/csharp/fsharp/perl-testing |
| Skill | ⭐ | Description |
|---|---|---|
python-patterns |
⭐⭐⭐⭐⭐ | Pythonic idioms, type annotations, best practices |
python-testing |
— | Merged into universal-testing (runtime: "Python") |
fastapi-patterns |
⭐⭐⭐⭐ | FastAPI + Pydantic v2, dependency injection, async routes |
django-patterns |
⭐⭐⭐⭐ | Django DRF, ORM, caching, signals |
django-celery |
⭐⭐⭐ | Django + Celery async tasks |
django-quality |
⭐⭐⭐ | Django quality suite (TDD + Security + Verification, 3-in-1) |
pytorch-patterns |
⭐⭐⭐ | PyTorch deep learning patterns |
| Skill | ⭐ | Description |
|---|---|---|
react-patterns |
⭐⭐⭐⭐⭐ | React 18/19 Hooks, server/client component boundaries |
frontend-patterns |
⭐⭐⭐⭐⭐ | Frontend architecture patterns (SKC + agent-skills) |
react-performance |
⭐⭐⭐⭐ | Vercel engineering practices; avoid common rendering pitfalls |
nextjs-turbopack |
⭐⭐⭐⭐ | Next.js 16+ / Turbopack incremental bundling |
vue-patterns |
⭐⭐⭐⭐ | Vue 3 Composition API, Pinia |
vite-patterns |
⭐⭐⭐⭐ | Vite config, plugins, HMR, SSR |
nestjs-patterns |
⭐⭐⭐ | NestJS modules, DTOs, guards, interceptors |
nuxt4-patterns |
⭐⭐⭐ | Nuxt 4 SSR, route rules, hydration |
angular-developer |
⭐⭐⭐ | Angular (includes 34 reference files) |
react-testing |
— | Merged into universal-testing (runtime: "React/TS") |
bun-runtime |
⭐⭐⭐ | Bun runtime / package manager / bundler / test runner |
backend-patterns |
⭐⭐⭐⭐ | Backend architecture patterns |
ui-to-vue |
⭐⭐⭐ | Batch-convert UI screenshots to Vue SFCs |
frontend-slides |
⭐⭐⭐ | Create/convert animated HTML presentations |
| Skill | ⭐ | Description |
|---|---|---|
kotlin-patterns |
⭐⭐⭐⭐ | Idiomatic Kotlin, coroutines, extension functions |
kotlin-testing |
— | Merged into universal-testing (runtime: "Kotlin") |
kotlin-coroutines-flows |
⭐⭐⭐ | Structured concurrency, StateFlow/SharedFlow |
kotlin-ktor-patterns |
⭐⭐⭐ | Ktor server routing DSL, plugins, authentication |
kotlin-exposed-patterns |
⭐⭐⭐ | JetBrains Exposed ORM |
springboot-patterns |
⭐⭐⭐ | Spring Boot REST, layered services, caching |
springboot-quality |
⭐⭐⭐ | Spring Boot quality suite (TDD + Security + Verification) |
quarkus-patterns |
⭐⭐⭐ | Quarkus 3.x LTS, Camel, RESTful |
quarkus-quality |
⭐⭐⭐ | Quarkus quality suite (TDD + Security + Verification) |
java-coding-standards |
— | Merged into coding-standards; standards moved to project .clauderules |
jpa-patterns |
⭐⭐⭐ | JPA/Hibernate entities, relationship mapping, query optimization |
tinystruct-patterns |
⭐ | TinyStruct Java framework guide |
| Skill | ⭐ | Description |
|---|---|---|
laravel-patterns |
⭐⭐⭐ | Laravel Eloquent, routing, queues |
laravel-quality |
⭐⭐⭐ | Laravel quality suite (TDD + Security + Verification) |
laravel-plugin-discovery |
⭐⭐ | Discover Laravel packages via LaraPlugins.io MCP |
| Skill | ⭐ | Description |
|---|---|---|
swiftui-patterns |
⭐⭐⭐ | SwiftUI @Observable, view composition, navigation |
swift-concurrency-6-2 |
⭐⭐⭐ | Swift 6.2 approachable concurrency |
swift-protocol-di-testing |
⭐⭐⭐ | Swift protocol-based dependency injection and testability |
swift-actor-persistence |
⭐⭐⭐ | Swift Actor thread-safe persistence |
foundation-models-on-device |
⭐⭐⭐ | Apple FoundationModels on-device LLM |
dart-flutter-patterns |
⭐⭐⭐ | Flutter null safety, state management |
flutter-dart-code-review |
⭐⭐⭐ | Flutter/Dart code review checklist |
android-clean-architecture |
⭐⭐⭐ | Android Clean Architecture, Hilt, ViewModel |
compose-multiplatform-patterns |
⭐⭐⭐ | Compose Multiplatform, KMP shared UI |
ios-icon-gen |
⭐⭐ | Generate iOS app icons from SF Symbols |
| Skill | ⭐ | Description |
|---|---|---|
dotnet-patterns |
⭐⭐⭐ | Idiomatic C# / .NET, dependency injection, async/await |
csharp-testing |
— | Merged into universal-testing (runtime: "C#") |
fsharp-testing |
⭐⭐ | F# testing: Unquote, FsCheck property testing |
| Skill | ⭐ | Description |
|---|---|---|
golang-patterns |
⭐⭐⭐⭐ | Idiomatic Go, concurrency, interfaces, error handling |
golang-testing |
— | Merged into universal-testing (runtime: "Go") |
rust-patterns |
⭐⭐⭐ | Ownership, error handling, traits, concurrency |
rust-testing |
— | Merged into universal-testing (runtime: "Rust") |
cpp-coding-standards |
— | Merged into coding-standards; standards moved to project .clauderules |
cpp-testing |
— | Merged into universal-testing (runtime: "C++") |
perl-patterns |
⭐⭐ | Modern Perl 5.36+ |
perl-testing |
— | Merged into universal-testing (runtime: "Perl") |
perl-security |
⭐⭐ | Perl taint mode, secure execution |
| Skill | ⭐ | Description |
|---|---|---|
postgres-patterns |
⭐⭐⭐⭐ | PostgreSQL query optimization, indexing, security |
redis-patterns |
⭐⭐⭐⭐ | Redis caching strategies, distributed locks, rate limiting |
docker-patterns |
⭐⭐⭐⭐ | Docker/Compose local dev, multi-stage builds |
kubernetes-patterns |
⭐⭐⭐ | K8s RBAC, probes, autoscaling |
database-migrations |
⭐⭐⭐ | Database migration best practices, zero downtime |
prisma-patterns |
⭐⭐⭐ | Prisma ORM schema, transactions, type safety |
mysql-patterns |
⭐⭐⭐ | MySQL queries, indexing, primary/replica replication |
clickhouse-io |
⭐⭐ | ClickHouse analytical query optimization |
latency-critical-systems |
⭐⭐⭐ | Low-latency system optimization patterns |
data-throughput-accelerator |
⭐⭐⭐ | Large-scale ETL throughput optimization |
| Skill | ⭐ | Description |
|---|---|---|
motion-design |
⭐⭐⭐⭐ | Complete animation collection (4-in-1): design tokens + production React patterns + advanced gestures + framework selection |
accessibility |
⭐⭐⭐⭐ | WCAG 2.2 Level AA accessibility design, implementation, and auditing |
frontend-a11y |
⭐⭐⭐ | Frontend accessibility best practices |
frontend-design-direction |
⭐⭐⭐ | Design direction and visual style guidelines |
make-interfaces-feel-better |
⭐⭐⭐ | Specific design details to make interfaces more polished |
liquid-glass-design |
⭐⭐ | iOS 26 Liquid Glass dynamic glass material |
Includes: agentic-engineering, autonomous-loops, agentic-os, cost-aware-llm-pipeline, agent-FRM-construction, MPX-workflows, team-agent-orchestration, team-builder, MAF, parallel-execution-optimizer, continuous-agent-loop, autonomous-agent-FRM, gan-style-FRM, DVR, recursive-decision-ledger, dynamic-workflow-mode, RFP-rfc-pipeline, iterative-retrieval, recsys-pipeline-architect, ml-adoption-playbook, mle-workflow
(See the previous README Agent/AI Engineering section for detailed descriptions)
orch-build-mvporch-add-featureorch-fix-defectorch-change-featureorch-refine-code
This round's refactoring: the 5 skills above have been merged into
master-orchestrator, which uses an intent-routing state machine (BUILD_MVP / ADD_FEATURE / FIX_DEFECT / CHANGE_FEATURE / REFINE_CODE) for automatic dispatch, sharing a single Research → Plan → TDD → Deliver pipeline.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
master-orchestrator |
⭐⭐⭐⭐ | orchestration | Unified orchestrator (5-in-1): intent-routing state machine auto-identifies BUILD_MVP / ADD_FEATURE / FIX_DEFECT / CHANGE_FEATURE / REFINE_CODE, drives a shared Research → Plan → TDD → Deliver pipeline with two confirmation gates (Gate 1: plan approval, Gate 2: commit confirmation). Replaces all 5 orch-* skills |
Purpose: Full validation suite before committing code — code review, automated testing, security scanning, health dashboards, agent evaluation.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
code-review |
⭐⭐⭐⭐⭐ | harness+skills+mattpocock+LRN (4-in-1) | Full code review lifecycle: Pre-PR self-review + initiate review + receive feedback + Pre-Landing checklist |
verification-before-completion |
⭐⭐⭐⭐ | skills+SKC (merged) | Multi-dimensional verification before claiming done: tests pass, edge cases, docs updated (includes SKC loop verification mechanism) |
automated-testing |
⭐⭐⭐⭐ | anthropic+LRN+SKC (3-in-1) | Automated test suite: Playwright functional/visual regression/accessibility + Lightpanda Web Vitals + SKC e2e POM |
code-health |
⭐⭐⭐⭐ | LRN (extracted) | Code quality dashboard: type-check, lint, test coverage, dead code, bundle size scoring; never auto-fixes |
improve-codebase-architecture |
⭐⭐⭐⭐ | mattpocock | Identify deep codebase issues (coupling, boundaries, naming); produce a prioritized refactoring list |
security-review |
⭐⭐⭐⭐ | SKC | Real-time security checks during development: auto-triggers when handling auth, user input, secrets, encryption, or authorization |
threat-model |
⭐⭐⭐⭐ | defending-code | Threat modeling (interview/bootstrap/hybrid three modes); outputs THREAT_MODEL.md |
vuln-scan |
⭐⭐⭐⭐ | defending-code | Static source code security scan; parallel sub-agents cover memory safety/injection/auth bypass |
vuln-triage |
⭐⭐⭐⭐ | defending-code | Security finding triage: multi-vote validation, deduplication, sort by exploitability; outputs concise TRIAGE.md |
patch |
⭐⭐⭐ | defending-code | Generate fixes for verified vulnerabilities: fix + independent review dual sub-agents |
security-scan |
⭐⭐⭐ | SKC | Scan .claude/ directory config: detect misconfigurations, overly broad permissions, secret leaks |
security-bounty-hunter |
⭐⭐⭐ | SKC | Hunt for bounty-eligible security vulnerabilities: focus on remotely exploitable, high-impact issues |
defi-amm-security |
⭐⭐ | SKC | Solidity AMM contract security checklist: reentrancy attacks, flash loans, slippage protection |
llm-trading-agent-security |
⭐⭐ | SKC | Security patterns for autonomous trading agents with wallet or trade authorization |
eval-FRM |
⭐⭐⭐ | SKC | Formal evaluation framework for Claude Code sessions: eval-driven development, session sampling, quality gates |
agent-eval |
⭐⭐⭐ | SKC | Agent comparative evaluation: Claude Code vs Aider vs Codex etc. on custom tasks |
agent-self-evaluation |
⭐⭐⭐ | SKC | 5-dimension post-task self-evaluation of agent output: accuracy, completeness, quality, efficiency, risk |
agent-architecture-audit |
⭐⭐⭐ | SKC | Full-stack agent/LLM application diagnostics: 12-layer agent stack audit |
agent-introspection-debugging |
⭐⭐⭐ | SKC | Structured AI agent failure self-debugging workflow: capture, diagnose, isolate |
ai-regression-testing |
⭐⭐⭐ | SKC | Regression testing strategy for AI-assisted development: sandbox-mode API testing |
ai-first-engineering |
⭐⭐⭐ | SKC | Engineering operations patterns for teams where AI agents generate large volumes of implementations |
benchmark |
⭐⭐⭐ | SKC | Measure performance baseline; detect regressions before/after PRs |
triage |
⭐⭐⭐ | mattpocock | Issue priority assessment: state-machine-driven triage, routes to the correct owner |
browser-testing-with-devtools |
⭐⭐⭐⭐ | agent-skills | Chrome DevTools MCP browser testing: runtime validation via Chrome DevTools Protocol — DOM state, network requests, Console errors, performance profiles; operate real browsers directly from Claude Code |
observability-and-instrumentation |
⭐⭐⭐⭐ | agent-skills | Observability and instrumentation: embed structured logging, RED metrics (rate/errors/latency), distributed tracing, symptom-based alerting at build time — "baked in, not bolted on" |
code-simplification |
⭐⭐⭐⭐ | agent-skills | Code simplification: reduce unnecessary complexity without changing behavior — identify over-abstraction, unprofitable indirection layers, inlineable single-line functions. Aligned with LZY philosophy |
LZY-review |
⭐⭐⭐ | LZY | Code review in LZY mode: examine PRs through the minimalist lens; identify "code that shouldn't exist"; suggest shortest working diff improvements |
browser-qa |
⭐⭐⭐ | SKC | Visual testing and UI interaction validation using browser automation |
click-path-audit |
⭐⭐⭐ | SKC | Trace the complete state-change chain for every user-facing button/touchpoint; find broken links |
ui-demo |
⭐⭐ | SKC | Record polished UI demo videos using Playwright |
windows-desktop-e2e |
⭐⭐ | SKC | E2E testing for Windows native desktop apps (WPF/WinForms/Win32) using pywinauto |
FCG |
⭐⭐⭐ | SKC | Fact-enforcement gate: blocks Edit/Write/Bash and requires specific evidence-based information as a mandatory audit |
production-audit |
⭐⭐⭐ | SKC | Evidence-based production readiness audit: pre-launch review and post-merge checks |
codehealth-mcp |
⭐⭐⭐ | SKC | Real-time structured code health via CodeScene MCP: review before editing, validate score delta |
WTQ-code-quality |
⭐⭐ | SKC | Write-time code quality enforcement using WTQ: auto-format, lint, Claude analysis integration |
prompt-optimizer |
⭐⭐⭐ | SKC | Prompt optimization tool: analyze and improve the effectiveness and efficiency of Claude prompts |
repo-scan |
⭐⭐⭐ | SKC | Cross-stack source asset audit: categorize every file, detect embedded third-party code |
design-system-review |
⭐⭐⭐ | SKC | Generate or audit design systems, check visual consistency, review Penpot exports |
Purpose: Push validated code to production — CI management, automated releases, deployment strategies, canary monitoring.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
completing-and-releasing |
⭐⭐⭐⭐⭐ | skills+harness+LRN (3-in-1) | Complete release workflow: validate tests → merge/PR → automated release (changelog + version bump + tag + GitHub Release) |
ci |
⭐⭐⭐⭐ | harness | CI pipeline failure triage: diagnose cause, propose fix, one-click recovery for red builds |
deployment-strategies |
⭐⭐⭐ | harness+LRN+SKC (3-in-1) | Three deployment strategies: Direct (Vercel/Netlify) + Canary (progressive traffic shift) + SKC detailed mode (Docker/K8s/blue-green) |
canary-watch |
⭐⭐⭐ | SKC | Post-release monitoring: check HTTP endpoints, detect error rates, validate response quality |
opensource-pipeline |
⭐⭐ | SKC | Open-source release pipeline: fork, strip private content, package into a safe public release |
Purpose: Post-release learning and maintenance loop — retrospectives, knowledge capture, maintenance, documentation, workspace audits.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
engineering-retro |
⭐⭐⭐ | openclaw+LRN (dual mode) | Engineering weekly retrospective (dual mode): git analysis edition (14 steps: commit metrics, hotspot files, per-person feedback) + conversational review edition |
code-documentation |
⭐⭐⭐⭐ | LRN (extracted) | Documentation dual mode: /document-generate to create from scratch, /document-release to update after launch |
maintenance |
⭐⭐⭐ | harness | Codebase housekeeping: clean up Plans.md, archive old session logs, remove expired state files |
FRM-sync |
⭐⭐ | harness | Sync Plans.md with git implementation state, detect drift, update progress markers |
workspace-surface-audit |
⭐⭐⭐ | SKC | Audit active repos, MCP servers, plugins, connectors, env surfaces, and harness configuration |
Purpose: Transform work outputs into shareable formats — office documents, videos, articles, social content.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
pdf |
⭐⭐⭐⭐⭐ | anthropic+LRN | Full PDF capability (read, merge, split, fill forms, OCR, watermark) + Lightpanda Markdown→PDF rendering |
docx |
⭐⭐⭐⭐ | anthropic | Word document creation and editing (TOC, headings, tables, page numbers, etc.) |
pptx |
⭐⭐⭐⭐ | anthropic | PowerPoint presentation creation and editing |
xlsx |
⭐⭐⭐⭐ | anthropic | Excel spreadsheets (formulas, charts, data cleaning) |
article-writing |
⭐⭐⭐⭐ | SKC | Write articles, tutorials, blog posts, press releases, and other long-form content: SEO optimization, structure design, tone adjustment |
visual-content |
⭐⭐⭐⭐ | baoyu+harness | SVG diagrams (architecture, flowcharts, sequence diagrams, mind maps) + project slides (Nano Banana Pro) |
brand-voice |
⭐⭐⭐ | SKC | Build source-based writing style profiles from real posts, articles, and documents |
content-engine |
⭐⭐⭐ | SKC | Multi-platform native content system: X, LinkedIn, TikTok, YouTube, Newsletter |
crosspost |
⭐⭐⭐ | SKC | Multi-platform content distribution: X, LinkedIn, Threads, Bluesky; adapted tone and format per platform |
x-api |
⭐⭐⭐ | SKC | X/Twitter API integration: post tweets, threads, read timeline, search, user analytics |
seo |
⭐⭐⭐ | SKC | SEO audit, planning, and implementation: technical SEO, on-page optimization, structured data |
doc-coauthoring |
⭐⭐⭐ | anthropic | Structured collaborative document workflow |
internal-comms |
⭐⭐⭐ | anthropic | Enterprise internal communication template library (emails, announcements, reports) |
markdown-docs |
⭐⭐⭐ | baoyu | Markdown documentation pipeline: standardize formatting → convert to themed HTML |
writing-craft |
⭐⭐ | mattpocock | Writing craft: editing and polishing + three-phase writing workflow (mine material → shape → narrative beats) |
video-editing |
⭐⭐⭐ | SKC | AI-assisted video editing: cut, structure, and enhance real footage |
manim-video |
⭐⭐ | SKC | Build reusable explanatory animations for technical concepts using Manim: diagrams, system charts, product demos |
remotion-video-creation |
⭐⭐ | SKC | Remotion best practices for React-based video creation: 29 domain rules |
videodb |
⭐⭐ | SKC | See/understand/act on video and audio: ingest local files, URLs, RTSP streams with AI understanding |
social-publisher |
⭐⭐ | SKC | Agent-driven social media publish scheduling across 13 platforms |
taste |
⭐ | SKC | Angelcore-style music video and short film creative direction layer |
nutrient-document-processing |
⭐⭐ | SKC | Process documents using Nutrient DWS API: convert, OCR, extract, edit, sign |
Purpose: Build AI tool extensions, web automation, and third-party service integrations.
| Skill | ⭐ | Source | Description |
|---|---|---|---|
ai-extensions |
⭐⭐⭐⭐⭐ | anthropic+mattpocock+skills+SKC (4-in-1) | Complete AI extension building collection: MCP server development + Skill creation (quick start + TDD eval + Benchmark) + SKC MCP patterns |
SKA |
⭐⭐⭐⭐ | SKA | Skill security audit tool: run before installing any unknown skill. Static analysis + LLM semantic detection of 14 threat categories (prompt injection, data exfiltration, MCP poisoning, etc.); outputs LOW/MEDIUM/HIGH/CRITICAL risk levels; includes SARIF output and CI integration |
web-access |
⭐⭐⭐ | harness+agent-reach | Web access dual mode: browser automation (Lightpanda CLI) + 17-platform content access (Twitter, Reddit, GitHub, etc.) |
github-ops |
⭐⭐⭐⭐ | SKC | GitHub repo automation: issue triage, PR management, CI/CD triggering |
google-workspace-ops |
⭐⭐⭐ | SKC | Unified cross-product workflow for Google Drive/Docs/Sheets/Slides |
email-ops |
⭐⭐⭐ | SKC | Evidence-first inbox triage, drafting, send verification, and follow-up workflow |
jira-integration |
⭐⭐⭐ | SKC | Jira ticket retrieval, requirements analysis, status updates, board operations |
project-flow-ops |
⭐⭐⭐ | SKC | Execution flow operations across GitHub and Linear: issue triage, PR linking, sprint management |
GWS-ops |
⭐⭐ | harness | Google Workspace CLI automation (Drive/Sheets/Docs/Slides) |
data-scraper-agent |
⭐⭐⭐ | SKC | Fully autonomous AI data collection agent: job boards, news, e-commerce, social media, and other public sources |
dashboard-builder |
⭐⭐⭐ | SKC | Build monitoring dashboards that answer real operational questions: Grafana, SigNoz |
fal-ai-media |
⭐⭐⭐ | SKC | Unified media generation via fal.ai MCP: images (SDXL/FLUX), video, audio |
enterprise-agent-ops |
⭐⭐⭐ | SKC | Operate long-running agent workloads: observability, security boundaries, lifecycle management |
unified-notifications-ops |
⭐⭐ | SKC | Unified notification management workflow across GitHub, Linear, and desktop alerts |
messages-ops |
⭐⭐ | SKC | Evidence-first instant messaging workflow: read texts or DMs, draft replies |
lead-intelligence |
⭐⭐ | SKC | AI-native lead intelligence and outreach pipeline; replaces Apollo, Clay, ZoomInfo |
connections-optimizer |
⭐⭐ | SKC | Reorganize X and LinkedIn social networks: prune, follow recommendations, network gap analysis |
social-graph-ranker |
⭐⭐ | SKC | Weighted social graph ranking: warm referral discovery, bridge scoring, network gap analysis |
customer-billing-ops |
⭐⭐ | SKC | Customer billing workflow: subscriptions, refunds, churn classification, collections |
finance-billing-ops |
⭐⭐ | SKC | Evidence-first workflow for revenue, pricing, refunds, and team billing |
agent-payment-x402 |
⭐⭐ | SKC | Add x402 payment enforcement to AI agents: per-task budgets, spend controls |
scientific-db-pubmed-database |
⭐⭐ | SKC | Direct PubMed/NCBI biomedical literature search workflow |
scientific-db-uspto-database |
⭐⭐ | SKC | USPTO patent and trademark data workflow |
scientific-pkg-gget |
⭐⭐ | SKC | gget CLI: fast genomics database queries, sequence retrieval, BLAST search |
flox-environments |
⭐⭐ | SKC | Create cross-platform reproducible development environments using Flox (macOS/Linux) |
uncloud |
⭐⭐ | SKC | Manage Uncloud clusters: deploy services, configure Caddy ingress |
blender-motion-state-inspection |
⭐ | SKC | Inspect Blender characters, bones, poses, and animation retargeting |
generating-python-installer |
⭐ | SKC | Commercial-grade Windows Python installer: Nuitka compilation, distribution packaging |
evm-token-decimals |
⭐ | SKC | Prevent decimal mismatch bugs on EVM chains: runtime precision lookup, cross-chain safe computation |
nodejs-keccak256 |
⭐ | SKC | Prevent Ethereum hashing bugs: difference between Node.js sha3-256 and NIST SHA3 |
terminal-ops |
⭐⭐ | SKC | Evidence-first repo execution workflow: safely run commands, verify outputs |
Purpose: UI/UX design exclusive layer, organized by content; not included in workflow-based merges. Target audience: Frontend developers and UI/UX designers, especially those using the React/Tailwind/shadcn stack.
| Skill | ⭐ | Description |
|---|---|---|
frontend-design |
⭐⭐⭐⭐⭐ | Production-grade frontend interfaces with bold aesthetics and high design quality |
ui-ux-pro-max |
⭐⭐⭐⭐ | Flagship UI/UX: 50+ styles, 161 color palettes, 57 font combinations, 99 UX guidelines, 10 tech stacks |
ui-styling |
⭐⭐⭐⭐ | shadcn/ui + Tailwind CSS component styling system |
detail-design |
⭐⭐⭐⭐ | UI micro-interactions and fine-grained detail (accessibility, buttons, browser compatibility) |
html-prototype |
⭐⭐⭐⭐ | Generate HTML/CSS prototypes from design descriptions; Lightpanda three-viewport screenshot validation |
design-workshop |
⭐⭐⭐⭐ | Design workshop: Consultation (deep exploration → DESIGN.md) + Shotgun (3–5 divergent concepts) |
visual-review |
⭐⭐⭐ | Visual UI review: screenshot-based inspection of spacing, color contrast, and type hierarchy |
design |
⭐⭐⭐ | General design decisions (web + mobile) |
design-system |
⭐⭐⭐ | Design system construction: token architecture, component specs, slide generation |
canvas-design |
⭐⭐⭐ | Programmatic graphic design using Python + Pillow |
brand-guidelines |
⭐⭐⭐ | Brand visual standards (color, typography, logo usage) |
theme-factory |
⭐⭐⭐ | Multi-theme system generation |
slides |
⭐⭐⭐ | Chart.js strategic presentations with responsive layouts |
brand |
⭐⭐ | Brand voice, visual identity, and asset management |
banner-design |
⭐⭐ | Multi-platform banner design |
algorithmic-art |
⭐⭐ | p5.js algorithmic art with seeded randomness |
FRM-ui |
⭐⭐ | UI component generation for the harness framework |
Purpose: Skills that deeply depend on the LRN toolchain (requires gbrain, browse daemon, preamble-tier).
Most LRN skills have already been extracted and migrated to their corresponding workflow layers; the remaining 15 are deeply dependent.
| Skill | Description |
|---|---|
autoplan |
Four-dimension review pipeline (CEO + Design + Engineering + DX) |
browse |
browse daemon headless browser wrapper |
codex |
OpenAI Codex CLI three-mode wrapper |
cso |
Chief Security Officer security audit |
devex-review |
Developer experience real-time audit |
LRN-overview |
LRN framework meta-skill |
investigate |
System debugging (LRN edition) |
ios-clean/fix/qa/sync/ios-design-review |
iOS development suite (5 skills) |
office-hours |
YC office hours (LRN edition) |
pair-agent |
Paired browser agent |
skillify |
Solidify scraped workflows into browser-skills |
Purpose: Highly specialized vertical domain skills, grouped by industry. Target audience: Developers and engineers working in specific industries.
| Skill | ⭐ | Description |
|---|---|---|
healthcare-cdss-patterns |
⭐⭐ | Clinical Decision Support System (CDSS) development patterns: drug interaction checks, dosage recommendations |
healthcare-emr-patterns |
⭐⭐ | EMR/EHR development patterns: clinical safety, visit workflows, HL7 FHIR integration |
healthcare-phi-compliance |
⭐⭐ | PHI/PII compliance: minimization principles, encryption, access controls, audit logging |
healthcare-eval-FRM |
⭐⭐ | Patient safety evaluation harness for healthcare applications: automated test suites |
hipaa-compliance |
⭐⭐ | HIPAA compliance entry point: privacy, security, and healthcare data protection |
| Skill | ⭐ | Description |
|---|---|---|
ito-basket-compare |
⭐ | Compare Itô prediction market baskets against user knowledge base and portfolio notes |
ito-data-atlas-agent |
⭐ | Backend Data Atlas-style agent design for Itô basket research |
ito-market-intelligence |
⭐ | Research Itô prediction market events, instruments, liquidity, and news context |
ito-trade-planner |
⭐ | Build non-advisory prediction market trade planning worksheets |
prediction-market-oracle-research |
⭐ | Research prediction markets as data sources or oracle signals for products/agents/dashboards |
prediction-market-risk-review |
⭐ | Compliance and security review for prediction markets, baskets, oracles, and trading agent workflows |
| Skill | ⭐ | Description |
|---|---|---|
homelab-network-readiness |
⭐⭐ | Homelab VLAN segmentation, local DNS filtering, and WireGuard-style VPN readiness checklist |
homelab-network-setup |
⭐⭐ | Home and homelab network planning: gateway, switches, APs, IP ranges |
homelab-pihole-dns |
⭐⭐ | Pi-hole installation, blocklist management, DNS-over-HTTPS, DHCP integration |
homelab-vlan-segmentation |
⭐⭐ | Segment home networks into IoT/Guest/Trusted/Server VLANs using UniFi |
homelab-wireguard-vpn |
⭐⭐ | WireGuard VPN server config, peers, key generation, split tunneling |
cisco-ios-patterns |
⭐⭐ | Cisco IOS/IOS-XE review patterns: show commands, config hierarchy, wildcard masks |
network-bgp-diagnostics |
⭐⭐ | Diagnostics-only BGP troubleshooting: neighbor state, route exchange, prefix filtering |
network-config-validation |
⭐⭐ | Pre-deployment checks for router and switch configurations: dangerous commands, duplicate addresses |
network-interface-health |
⭐⭐ | Diagnose interface errors, drops, CRC errors, duplex mismatches, and flapping |
netmiko-ssh-automation |
⭐⭐ | Secure Python Netmiko read-only collection, bulk SSH patterns, TextFSM parsing |
| Skill | ⭐ | Description |
|---|---|---|
carrier-relationship-management |
⭐ | Carrier relationship management: negotiation strategy, performance tracking, contract management |
customs-trade-compliance |
⭐ | Customs and trade compliance: HS codes, import/export regulations, trade documentation |
energy-procurement |
⭐ | Energy procurement: market analysis, contract negotiation, sustainable energy strategy |
inventory-demand-planning |
⭐ | Inventory and demand planning: demand forecasting, safety stock calculation, replenishment optimization |
logistics-exception-management |
⭐ | Logistics exception management: workflows for handling delays, damage, and lost shipments |
production-scheduling |
⭐ | Production scheduling: schedule optimization, bottleneck identification, resource allocation |
quality-nonconformance |
⭐ | Quality nonconformance management: root cause analysis, corrective actions, preventive actions |
returns-reverse-logistics |
⭐ | Returns and reverse logistics: return processing, refurbishment, and recycling workflows |
See CONTRIBUTING.md (English) in the root directory for full specifications on adding or modifying skills. Core constraints:
- Single
SKILL.mdmust not exceed 4,000 tokens (~3,000 words / 16,000 characters). Exceeding this limit significantly degrades the model's instruction-following for the latter half of the file. Oversized files should be split into a dispatchSKILL.md+modes//references//domains/subdirectories. - All files (except this README) must be in English.
- Do not place
README.md,INDEX.md, orLICENSE.txtinside skill folders. - Shared assets go in
_shared/directories; cross-skill duplication is not allowed.
Size savings: approximately −8.5 MB
| Optimization | Action | Savings |
|---|---|---|
| Font deduplication | Merged canvas-fonts/ from 09-design/canvas-design/ and ui-styling/ into 09-design/_shared/canvas-fonts/ |
−5.5 MB |
| Office script/schema deduplication | Merged identical XSD schemas, soffice.py, helpers/, pack.py, unpack.py, validate.py from docx / pptx / xlsx into 07-share/_office-shared/ |
−2.2 MB |
| Invalid file cleanup | Deleted all 11 LICENSE.txt files, 18 INDEX.md files, and 30 non-root README.md files |
−0.3 MB |
Oversized SKILL.md splits (4 files, reducing ~780 KB of scattered loading):
| Skill | Original Size | Split Method |
|---|---|---|
02-define/plan-review |
314 KB | → dispatch SKILL.md + modes/ (ceo/eng/mega/design/devex five mode files) |
11-vertical/business-ops/supply-chain-ops |
199 KB | → dispatch SKILL.md + domains/ (8 domain files) |
04-assure/automated-testing |
138 KB | → dispatch SKILL.md + references/ (playwright guide + qa-bootstrap guide) |
05-release/release-pipeline |
132 KB | → dispatch SKILL.md + references/ (completing-releasing / deployment-strategies / canary-watch) |
Skill merges and ghost cleanup (−10):
| Action | Details |
|---|---|
Merged into autonomous-loops |
continuous-agent-loop (file self-declares supersedes) |
Merged into FRM-design |
agent-FRM-construction + dynamic-workflow-mode (files declare Replaces) |
Merged into team-orchestration |
team-builder + team-agent-orchestration (files declare Replaces) |
Merged into ml-engineering |
ml-adoption-playbook + mle-workflow (files declare Replaces) |
Merged into release-pipeline |
completing-and-releasing + canary-watch + deployment-strategies (files declare Replaces) |
Other fixes:
- Fixed
prompt-optimizerempty description (original""caused it to never trigger) - Added missing
tagsfield tomessy-archive-curator - Fixed inverted logic comment in
autonomous-loops(original mistakenly namedcontinuous-agent-loopas the canonical name) - Added clear negative trigger words to 4 agent evaluation skills (
agent-eval/agent-self-evaluation/agent-architecture-audit/agent-introspection-debugging) to eliminate ambiguity - Fixed 11 cross-file dead references (
/canary-watch→/release-pipeline, etc.) - Unified redundant
automationtags in 3108-extendskills toextend
v6 round additions (this round):
- Cleared 60 ghost skills of the same type (covering all unexecuted Replaces declarations in 23 successor skills) across layers:
01-discover,02-define,03-build(python/php/jvm/mobile/data-infra/ui-motion),04-assure,06-reflect,07-share,08-extend,11-vertical(finance/network) - Fixed
vuln-triage/SKILL.mdname collision (name field was mistakenlytriage, clashing with04-assure/triage) - Fixed
04-assure/design-system-reviewname field (was mistakenlydesign-system, clashing with09-design/design-system) - Fixed incorrect tags on
djangoandlaravel(jvm→python/php) - Removed
ckm:namespace prefix from 6 skills under09-design(internal framework remnant, zero external references) - Standardized
Replaces:in descriptions of 25 successor skills toIncorporates former:
| Source | Description |
|---|---|
| anthropic | General skills officially released by Anthropic |
| SKC | SKC — large skill collection covering 30+ language frameworks |
| harness | FRM enterprise-grade multi-agent framework |
| LRN (extracted) | Skills extracted from the LRN framework (gbrain/preamble/browse dependencies replaced) |
| skills | Parallel agent and Git worktree development philosophy |
| mattpocock | Matt Pocock TypeScript ecosystem engineering practices |
| openclaw | LRN openclaw general skills (extracted) |
| baoyu | Diagram and Markdown processing |
| agent-reach | Multi-channel content access across 17 platforms |
| pm-brain | PM second brain system |
| defending-code | Anthropic open-source security vulnerability analysis framework |
| SKA | SKILL.md file security audit tool (includes SKILL.md and companion scripts/) |
Target audience: Developers, product managers, designers, or any technical/non-technical job seekers.
career-ops is a complete AI-driven job search management system supporting 20 operation modes, covering everything from resume generation to negotiation strategy.
| Skill | ⭐ | Description |
|---|---|---|
career-ops |
⭐⭐⭐⭐ | Complete job search management system (20 modes): ATS keyword matching and scoring, CV generation (HTML/PDF/LaTeX three formats), cover letter writing, in-depth company research, automated application pipeline, interview preparation, follow-up emails, negotiation strategy, LinkedIn optimization, salary benchmarking, application tracking pipeline management, etc. Supports Playwright PDF rendering, CSV/XLSX batch processing, multi-ATS integration (Greenhouse/Lever/Workday) |
career-ops companion content:
| Subdirectory | Contents | Description |
|---|---|---|
modes/ |
20 operation mode Markdown files | Complete instructions for each scenario (CV, cover-letter, research, etc.) |
templates/ |
HTML/LaTeX/YML templates (6 files) | Professional layout templates for CVs and cover letters with PDF export |
config/ |
profile.example.yml | Candidate profile configuration example |
scripts/ |
28 .mjs automation scripts | Playwright PDF generation, ATS scanning, tracking sheet management, batch applications, etc. |
providers/ |
ATS data providers | Data retrieval layer for Greenhouse/Lever/Workday and other platforms |
examples/ |
Sample CVs and profiles | Reference implementations showing output format after full configuration |
scaffolder/ |
CLI scaffolding tool | Quickly initialize a new career-ops workspace |
batch/ |
Batch processing documentation | Configuration and prompt documentation for batch application mode |
fonts/ |
Font files | Font resources required by PDF/HTML templates |
Prerequisites: Node.js 18+,
npm install(in thescripts/directory), optional Playwright (for PDF export)
All merge operations use only copy, delete, and concatenation — no semantic rewriting (merges other than Rule 5 are append-only).