700+ free AI skills. Pick one.
Build today.
The complete 2BFT skill library. Browse, filter, start. No paywall, no signup.Each skill is one specific thing you can learn in 15 minutes to 1 day — not a 40-hour course you’ll never finish.
Skills, not courses. Pick one.
Every entry below is a single, specific thing you can do with AI — not a 12-week bootcamp. Open the category that matches what you’re trying to ship, pick a skill that fits your level, and build it. Most take 15 minutes to 1 day. All of them are free.
We update this list every month. The 700+ total includes deep-dive skills in each category — what you see here is the 78 most-taught. The full library is split across /learn (200+ resources) and /agentic-systems (100+ copy-ready workflows).
Prompt Engineering
The skill every other skill rests on. How to talk to a model so it actually does what you meant.
Write a system prompt in 3 sentences
Few-shot a model into your brand voice
Force structured JSON output with tool use
Debug a prompt with chain-of-thought + evals
Translate prompts between GPT-5.6, Claude Sonnet 5, Gemini 3.1 Pro
Build a 9-section system prompt that doesn't drift
Write a prompt that refuses gracefully
Use temperature 0 for evals, 0.7 for brainstorming
AI-Assisted Coding
Ship code with Claude Code, Cursor, Codex, Trae. The new IDE is a chat box.
Set up Claude Code in 10 minutes
Cursor rules that ship every project
Write a CLAUDE.md that actually helps
Use sub-agents to parallelise a feature
Debug a stack trace with Claude Code
Build a Next.js app end-to-end with v0 + Cursor
Ship a Vibe Coding Hackathon project in 48 hours
Visual AI — Nano Banana 2
Image and video, the 2026 stack. Nano Banana 2 for stills, Veo 3.1 for video.
Generate a 4K product shot with Nano Banana 2
Swap a model's face while keeping the lighting
Build a 30-image catalogue shoot in 1 hour
Generate a Veo 3.1 8-second ad with audio
Make a consistent character across 12 frames
Replace a background without a green screen
Batch-generate 100 ad creatives from a CSV
Voice AI — ElevenLabs
Audio content at scale. Voiceovers, podcasts, narration, real-time voice agents.
Clone your own voice in ElevenLabs (5 min setup)
Dub a 5-minute YouTube video into Tamil
Build a 24/7 voice agent for your shop
Make a podcast with two AI hosts
Generate an audio ad that sounds human
Add real-time voice to a WhatsApp bot
Workflow Automation
n8n, Make, Zapier, OpenClaw. Stop doing the same 5 things every day by hand.
Set up a WhatsApp Business API agent
Auto-post to Instagram from a Notion calendar
Build an n8n flow that handles 3 branches
Replace 5 daily Zaps with one OpenClaw agent
Auto-reply to every DM with a custom prompt
Back up your SaaS data to Google Sheets nightly
Agentic AI
From a single prompt to a multi-agent system. OpenClaw default, LangGraph for state, Mastra for TS.
Install OpenClaw (210K+ stars) on a $5 VPS
Wire up an agent that uses 3 tools
Build a multi-agent crew with CrewAI
Ship a Mastra agent in TypeScript
Add a critic agent that reviews another agent's output
Self-host an agent on a Raspberry Pi
Build a daily-runner agent: scrape → summarise → email
RAG & Vector DBs
Teach a model your data. Retrieval-Augmented Generation, from prototype to enterprise.
Build a RAG pipeline over a Notion workspace
Chunk a 200-page PDF the right way
Pick a vector DB: Pinecone, Weaviate, Qdrant, pgvector
Add hybrid search (keyword + semantic)
Eval your RAG: faithfulness, relevance, context recall
Build a RAG over WhatsApp chat history
Multimodal AI
Models that see, hear, and read. Gemini 3.1 Pro, GPT-5.6, Claude Sonnet 5 — text + image + audio in one call.
Send an image to Claude Sonnet 5 in one call
Extract structured data from a bill photo
Build a 2M-context doc Q&A with Gemini 3.1 Pro
Read a hand-written Tamil prescription
Generate audio + image + text in one prompt
Use multimodal for accessibility on your site
MCP — Model Context Protocol
The missing protocol. The standard for letting models talk to your tools, files, and APIs.
Set up an MCP server for your files in 15 minutes
Wire Claude Code to GitHub via MCP
Build an MCP server for Notion in Python
Expose a database as MCP tools
Defend against prompt injection in MCP tools
Use 5 pre-built MCP servers in one Claude session
Indian AI Tools
Made in India, for India. Zia LLM, Sarvam, Krutrim, IndiaAI, and the Hindi/Tamil ecosystem.
Try Zoho's Zia LLM in production
Build with Bhashini's 22 Indian language APIs
Use Sarvam-1 for Tamil code-mixed chat
Apply for an IndiaAI compute grant (₹10,372 Cr)
Fine-tune Krutrim-2 on your own data
Ship a Hindi content engine with GPT-5.6 + Bhashini
AI for SMBs
Jewellers, D2C brands, restaurants, salons. Built for tier-3 towns, real Indian shopkeepers.
Build a WhatsApp catalogue with Nano Banana 2
Set up a daily Instagram post in 15 minutes
Auto-reply to every WhatsApp enquiry
Build a jewellery look-book for Diwali
Make a 30-second shop ad for ₹0
Track competitors' prices with an agent
Set up Maruthi Jewellers-style product photos
AI for Schools
Teachers, students, parents. Khanmigo, NotebookLM, MagicSchool AI. Free, practical, India-ready.
Set up Khanmigo as a free AI tutor for your child
Use NotebookLM to summarise the syllabus in 10 min
Build a custom Tamil-medium lesson plan with Claude
Generate 50 practice questions for Class 10 Maths
Make a video explainer for a hard concept in 10 min
Build a school WhatsApp bot for parent queries
You picked a skill. Now find the resource.
200+ free tutorials, docs, repos, and YouTube channels — curated, tagged by difficulty, updated monthly. Or grab a copy-ready agentic system and ship in a day.