AI Skills & Tools

A curated, practical library of AI tools, agent skills, and workflow tips.

Use this repo as a browseable reference. Each directory contains a concise profile that explains when to use a tool or skill, what it is good for, practical tips, and common watch outs.

---

Browse By Aim

Each row is a directory in this repo. Click the name for the human-readable profile.

Write code

Skill	Best used for
github-copilot	GitHub-native coding assistance and repository guidance
cursor	Deep codebase understanding, scoped rules, and model choice
continue	Local or private coding assistance with flexible model providers
aider	Terminal-native, git-integrated batch refactors
windsurf	Autonomous multi-step coding inside an IDE
caveman	Terse multi-agent output and lower token usage
claude-mem	Persistent cross-session memory for coding agents
claude-code	Terminal-native autonomous coding with repo context
codex	OpenAI-native coding agent workflows
cline	Open-source VS Code agent with MCP and tool approvals
roo-code	Configurable open-source VS Code coding agent modes
opencode	Open terminal coding agent with model flexibility
kiro	Spec-driven agentic development environment

Methodology skills

Skill	Best used for
brainstorming	Clarifying intent, scope, constraints, and success criteria before creative work
verification-before-completion	Requiring evidence before claiming work is complete
tdd	Driving features and fixes through red-green-refactor
systematic-debugging	Reproducing, isolating, root-causing, and defending bug fixes
writing-plans	Turning a confirmed brief into a phased implementation plan
executing-plans	Working through a written plan with checkpoints
dispatching-parallel-agents	Splitting independent work across parallel agents
using-git-worktrees	Isolating feature work in separate worktrees

Build agents and automations

Skill	Best used for
langchain	LCEL composition, modern tool binding, and RAG patterns
langgraph	Stateful graphs, checkpointers, and human-in-the-loop interrupts
crewai	Role-based agents, tasks, memory, and tools
autogpt	Agent Protocol REST and Forge SDK for custom abilities
n8n	Webhook triggers, native AI nodes, and custom node packaging
llamaindex	Data-grounded RAG and document-agent applications
pydantic-ai	Typed Python agents and structured outputs
openai-agents-sdk	OpenAI-native agents, tools, handoffs, tracing, and guardrails
anthropic-agent-sdk	Claude-centered agent SDK patterns
autogen	Microsoft multi-agent orchestration and research workflows
semantic-kernel	Microsoft AI orchestration for .NET, Python, and Java
mastra	TypeScript agents, workflows, RAG, and evals
vercel-ai-sdk	Streaming AI features and tool calling for TypeScript web apps
dspy	Evaluation-driven LLM program optimization

Evaluate and observe AI systems

Skill	Best used for
braintrust	CI/CD evals, production traces, and model/prompt comparisons
phoenix	Open-source AI observability, tracing, and evals
langfuse	Self-hostable LLM tracing, prompt management, and scores
weave	W&B-native LLM tracing and experiment lineage
deepeval	pytest-style LLM, RAG, and agent evaluation
ragas	RAG quality metrics and retrieval evaluation
promptfoo	Prompt, model, and red-team regression tests

Connect tools and context

Skill	Best used for
mcp	Standardizing agent access to tools, resources, and prompts
fastmcp	Building production-friendly Python MCP servers
modelcontextprotocol-servers	Reusing common MCP server integrations

Run and retrieve

Skill	Best used for
litellm	Model routing, budgets, fallbacks, and provider abstraction
ollama	Local open-model development and private prototypes
vllm	High-throughput open-model inference serving
qdrant	Production vector search and RAG retrieval
chroma	Local-first vector storage and RAG prototypes
pgvector	PostgreSQL-native vector search

Generate content

Skill	Best used for
jasper	Brand-voiced campaigns and enterprise marketing content
copy-ai	GTM workflows and personalised outreach pipelines
writesonic	SEO articles, Chatsonic, and Botsonic workflows
perplexity	Cited, web-grounded research and answers
notion-ai	Notion workspace writing, summaries, and lightweight RAG

Work with data

Skill	Best used for
julius-ai	Conversational data analysis from CSVs and live database connectors
akkio	Predictive analytics with confidence-aware outputs
obviously-ai	Explainable no-code predictions
bardeen	Browser automation, scraping, and enrichment workflows
polymer	Dashboards, data flow, and shareable embeds

Detailed comparisons, pricing notes, and audience quick-reference tables live in INDEX.md.

---

Adding Or Updating An Entry

1. Create or edit a top-level directory such as langchain/ or tdd/.
2. Keep the entry in one README.md.
3. Include category metadata near the top using the existing blockquote style.
4. Add practical sections: When To Use, Practical Tips, and Watch Outs.
5. Add or update the matching row in INDEX.md.

Superpowers Contributor Workflow

Superpowers is documented as a recommended way to work on this repository, not as a product feature of the catalog.

Recommended workflows:

• Use brainstorming before adding a new tool, skill, or tip category.
• Use writing-plans for multi-entry restructures or taxonomy changes.
• Use verification-before-completion before claiming a catalog update is done.
• Use systematic-debugging only when a script, validation command, or generated artifact fails.

Repo Guides

• INDEX.md — categorised tool index with pricing, audience, and recommendation notes.
• docs/CONTRIBUTING.md — contribution checklist for library entries.
• AGENTS.md — operational guidance for Codex sessions in this repo.
• CLAUDE.md — operational guidance for Claude Code sessions in this repo.

---

Purpose Green · AI Skills & Tools Library · 2026