Open agent-skill format
425 concepts across 12 domains of artificial intelligence, packaged as skills any agent can load. Written for an adult with no computer science and no mathematics past school arithmetic — so your agent teaches from what is written here, not from whatever it happens to recall.
Copy the domains you want into your agent’s skills folder. Take one or take all 12 — each stands on its own.
git clone https://github.com/basava-code/learn-ai-skills.git
# Claude Code
cp -r learn-ai-skills/mathematical-foundations ~/.claude/skills/
# Hermes
cp -r learn-ai-skills/mathematical-foundations ~/.hermes/skills/
Every concept names the graph in both directions, so an agent can check you are ready for something before it teaches it — and tell you what opens up once you are.
What has to come first. Eigenvalues and Eigenvectors names three:
Linear Transformations Matrix Multiplication Vector Spaces and RankWhat opens once it is understood. Vector Spaces and Rank names ten:
Eigenvalues and Eigenvectors PageRank Singular Value Decomposition Support Vector Machines and six moreOne skill each: a short index plus one Markdown file per concept, loaded on demand. Your agent reads 12 one-line descriptions at startup and opens a concept only when it needs it.
83 concepts
Fine-Tuning, Hallucination Mitigation, Inference Optimization and 5 more
77 concepts
Bayesian Methods, Calculus, Discrete Mathematics and 8 more
55 concepts
Backpropagation, Computer Vision, Convolutional Networks and 5 more
54 concepts
Algorithms and Complexity, Causal Inference, Data Structures and 7 more
35 concepts
Software Architecture, Software Testing, Spec-Driven Development
27 concepts
Data Engineering, Data Governance, Data Quality and 1 more
23 concepts
Cloud Computing, MLOps, Self-Hosting
22 concepts
AI and Society, Entrepreneurship, Talent Management
18 concepts
AI Agents, AI Coding Assistants, Model Context Protocol
15 concepts
AI Security, Cryptography, Privacy Engineering
9 concepts
AI in Finance, Biomedical Data Science
7 concepts
Fairness and Bias, Model Interpretability