ai_skills
AI skills and workflow templates for Claude Code, Copilot, Gemini, any AI tool that will ingest skills files.
AI Skills — 35+ Deep Domain Skills for Claude Code, Cursor, Codex & Beyond
Not 1,000 shallow templates. 35 skills that actually teach your AI how to think about a domain. Each skill is 200-500 lines of battle-tested patterns, real code examples, decision frameworks, and troubleshooting guides — the kind of knowledge that turns a generic LLM into a domain expert. (AKA - I've learned from my mistakes and repetition)
Why This Collection?
Most skill repos give you a paragraph per topic. This one gives you an engineering reference per topic. The difference:
| Approach | Depth | Result |
|---|---|---|
| Typical skill repo | "Use Docker. Pin image tags. Set restart policies." | LLM knows keywords but not tradeoffs |
| This repo | Full Docker Compose templates, TrueNAS-specific patterns, GPU passthrough, reverse proxy integration, migration workflows, troubleshooting tables | LLM generates production-ready stacks and debugs real problems |
Every skill explains why, not just what. The LLM reads these and understands the reasoning behind decisions — so it can adapt to your specific situation instead of parroting templates.
Quick Install
# Full collection
git clone https://github.com/drewid74/ai_skills.git ~/.claude/skills/ai_skills
# Or cherry-pick individual skills
cp -r ai_skills/docker-selfhost ~/.claude/skills/
Works with Claude Code, Cursor, Codex CLI, Windsurf, and any tool that reads SKILL.md files.
Skills by Category
Infrastructure & Homelab
| Skill | Lines | What It Does |
|---|---|---|
| docker-selfhost | 134 | Docker Compose generation, TrueNAS SCALE, reverse proxy, self-hosted service stacks |
| truenas-ops | 355 | ZFS, TrueNAS API scripting, dataset management, migration, replication, backup/restore |
| proxmox-k3s-infra | 194 | Proxmox VE, VMs, LXC, GPU passthrough, K3s clusters, Helm, GitOps with FluxCD/ArgoCD |
| deploy-pipeline | 423 | SSH/rsync deploy scripts (bash + PowerShell), secrets management, Dockge, rollback strategies |
| infrastructure-as-code | 265 | Terraform/OpenTofu, Ansible, state management, modules, roles, Terraform+Ansible handoff |
| llm-inference-stack | 234 | Ollama, vLLM, NVIDIA NIM, LiteLLM routing, VRAM sizing, quantization, multi-node topology |
| service-integration | 360 | n8n, Node-RED, message queues, notification pipelines, Traefik, Uptime Kuma, cron patterns |
Software Engineering
| Skill | Lines | What It Does |
|---|---|---|
| full-sdlc | 242 | Requirements through production — scaffolding, branching, testing, CI/CD, release management |
| code-reviewer | 205 | Systematic review: correctness, security, edge cases, performance, Python/JS/Go/Bash patterns |
| cicd-pipeline | 344 | GitHub/Forgejo Actions, self-hosted runners, build caching, container registries, release automation |
| github-workflow | 144 | Repo scaffolding, branch strategies, PR management, GitHub Actions, release workflows |
| testing-framework | 374 | pytest, Jest/Vitest, Playwright, k6 load testing, mocking, fixtures, CI integration, debugging flaky tests |
| database-architecture | 202 | Schema design, indexing, EXPLAIN ANALYZE, ORMs, migrations, connection pooling, replication, PostgreSQL ops |
| api-integration | 502 | REST/GraphQL/WebSocket design, OAuth/JWT, resilience patterns (retry, circuit breaker), webhooks |
AI & Machine Learning
| Skill | Lines | What It Does |
|---|---|---|
| agentic-architecture | 279 | Agent design patterns (ReAct, multi-agent), RAG pipelines, tool use, guardrails, evaluation |
| mcp-server-dev | 345 | Build MCP servers (FastMCP + TypeScript), tool design, transports, Docker packaging, federation |
| training-pipeline | 237 | LoRA/QLoRA fine-tuning, data prep, hyperparameters, DeepSpeed, autonomous training loops, MLflow |
| federated-memory | 212 | Agent memory (working/recall/archival), vector stores, federation model, sync, graph knowledge |
| ai-skills-dev | 175 | Skill development lifecycle, prompt engineering, agent architecture, cross-platform portability |
Security & Operations
| Skill | Lines | What It Does |
|---|---|---|
| security-reviewer | 314 | OWASP Top 10, container hardening, SSH/firewall, TLS, secrets management, scanning pipelines |
| observability-sre | 478 | Prometheus, Grafana, Loki, OpenTelemetry, alerting philosophy, SLI/SLO, incident response, runbooks |
Data & Intelligence
| Skill | Lines | What It Does |
|---|---|---|
| data-engineering | 282 | ETL/ELT pipelines, pandas/polars/DuckDB, data validation, file formats, cleaning, orchestration |
| sigint-osint-feeds | 280 | APRS, ADS-B, USGS, NOAA, GDELT, RSS aggregation, satellite tracking, PostGIS, worker patterns |
| archivebox-knowledge | 172 | Web archival, Paperless-NGX, content extraction, summarization pipelines, knowledge base integration |
| deep_research | 29 | Source-grounded web research with triangulation, confidence scoring, and citation |
Web & Frontend
| Skill | Lines | What It Does |
|---|---|---|
| web-performance-a11y | 193 | Core Web Vitals, asset optimization, WCAG 2.1/2.2 AA compliance, Lighthouse, SEO |
| frontend_design_ux_enforcement.md | 34 | Design systems, Tailwind v4, semantic HTML, accessibility audits, mobile-first patterns |
| browser-automation | 273 | Playwright, scraping patterns, anti-bot handling, e2e testing, page monitoring |
Productivity & Communication
| Skill | Lines | What It Does |
|---|---|---|
| content-strategy | 193 | Technical writing, READMEs, blog posts, API docs, SEO, email, documentation systems |
| productivity-automation | 156 | Cron, batch workflows, data transformation, monitoring, backup automation, templates |
| google_workspace_assistant | 32 | Gmail, Calendar, Sheets, Drive automation with drafting-first workflow |
Meta / Reasoning
| Skill | Lines | What It Does |
|---|---|---|
| project_orchestrator | 46 | Multi-agent orchestration — DAG task decomposition, delegation, checkpointing |
| sequential_thinking | 33 | Structured reasoning trees, root cause analysis, architecture review, decision matrices |
| repo_auditor | 47 | Repository health audit — structure, docs, links, dependencies, AI-readiness |
| ham-radio-network | 140 | Antenna math, CHIRP CSV, DMR codeplug, FT8/APRS, AREDN mesh, VLAN/firewall design |
What Makes These Different
Depth over breadth. Each skill is a focused engineering reference, not a feature list. Here's what you get inside a typical skill:
- Decision frameworks ("when to use X vs Y, and why")
- Practical code examples (Python, TypeScript, bash, YAML) you can use immediately
- Architecture patterns with tradeoff analysis
- Configuration templates with explanations
- Troubleshooting tables mapping symptoms to fixes
- Tool comparisons with honest assessments
Universal, not personal. Every skill uses placeholders (<NAS_IP>, <API_KEY>, <POOL_NAME>) and explains patterns that work across any setup. No hardcoded infrastructure.
Principles over procedures. Instead of "here's how to configure Stripe," you get "here's how to build resilient API integrations" — so the LLM can apply the pattern to any service.
Skill Anatomy
skill-name/
├── SKILL.md # Required: YAML frontmatter + markdown instructions
├── scripts/ # Optional: helper scripts for deterministic tasks
├── references/ # Optional: supplemental documentation
└── assets/ # Optional: templates, configs, icons
The YAML frontmatter controls when the skill activates:
---
name: docker-selfhost
description: "Use this skill whenever the user wants to work with Docker,
Docker Compose, containers, TrueNAS Scale, self-hosted services..."
---
The markdown body is the knowledge the LLM reads when the skill triggers. Write it like you're briefing a smart colleague who needs context, not step-by-step hand-holding.
Cross-Platform Reference Catalogs
Bonus: capabilities catalogs for comparing what each AI platform can do.
| Catalog | Description |
|---|---|
| claude_capabilities_catalog.md | All available tools, MCPs, skills, and triggers in Claude |
| chatgpt_capabilities_catalog.md | ChatGPT equivalent capabilities |
| gemini_capabilities_catalog.md | Gemini capabilities |
| cross_agent_skills.md | Audit prompt to generate a comparable catalog for any platform |
Installation
Full Collection
git clone https://github.com/drewid74/ai_skills.git ~/.claude/skills/ai_skills
Cherry-Pick Skills
# Copy just the skills you want
cp -r ai_skills/docker-selfhost ~/.claude/skills/
cp -r ai_skills/security-reviewer ~/.claude/skills/
Verify
Skills appear automatically once they're in your skills directory. The LLM reads the frontmatter description field to decide when to invoke each skill.
Skill directory paths vary by tool. For Claude Code, the default is
~/.claude/skills/. Check your tool's documentation for the correct location.
Contributing
Skills are plain markdown. The barrier to entry is low:
- Fork the repo
- Create a directory with a
SKILL.mdfile (use ai-skills-dev for the template) - Write 200-500 lines of domain knowledge with code examples and troubleshooting
- Open a PR with 2-3 example prompts that validate your trigger description
Quality bar: if a developer with 5 years of experience would learn something from your skill, it belongs here. If it's just paraphrasing documentation, it doesn't.
License
MIT — use these skills however you want, commercially or otherwise.
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found