This repository collects my personal AI notes, agents, skills, and a small personal library of utilities and experiments.
- Overview: quick orientation to this workspace.
- Notes & References: curated notes such as Claude experiments and other writeups.
- Agents & Skills: useful agent definitions, prompt patterns, and skill implementations.
- Personal library: small helpers, datasets, and reusable code.
Open this repository in your editor and explore the folders below. The repo is intentionally lightweight — add notes, agents, and small tools as you build them.
andrej-karpathy/— notes and writeups (example: Andrej-style learning notes).agents.md— (recommended) central index of agent configs, prompts, and usage examples.skills/— (recommended) folder for skill implementations and small utilities.lib/— (recommended) personal library: reusable functions, scripts, and data helpers.
agents.md— catalog of agent prompts, roles, and example invocations.skills/— store skill definitions and small modules used by agents.lib/— small Python/JS modules or notebooks that you reuse across projects.
- Keep each agent entry short: purpose, invocation example, and key prompts.
- For skills, include a README in
skills/describing interfaces and example usage. - Use clear naming like
skill_text_summary.pyorskill_data_cleaner.js.
- Organize utilities by language (e.g.,
lib/python/,lib/js/). - Add small README files describing APIs and simple examples.
Add new notes, agents, or skills as you discover them. Keep entries minimal and link back to original sources where appropriate.
- Create
agents.mdand populate with your most-used prompts. - Add a
skills/README.mddescribing interfaces. - Start organizing reusable code into
lib/.
If you want, I can scaffold agents.md, skills/README.md, and a starter lib/ layout now.