skills

A collection of skills built from research notes.

Each skill is grounded in a documented method or technique. The research and rationale behind every skill lives in my Obsidian vault. This repo holds the executable skills themselves.

Structure

skills/
  <skill-name>/
    SKILL.md      # the skill definition (frontmatter + instructions)

Skills

Skill	Category	What it does
learning-worked-example	learning	Tutors a topic using the Worked Examples method — atomic, step-by-step breakdowns for better retention.
diet-coach	health/fitness	Evidence-based, dynamic diet & weight-loss coach — reads a personal profile + progress log each run and recomputes calories, protein, phase (grind/chill) and steps. Answers in the user's language.
code-kiss	engineering	Keep It Simple — biases code toward the simplest solution that solves the actual problem; flags over-engineering, premature flexibility, and needless cleverness. Stack-agnostic.
code-dry	engineering	Don't Repeat Yourself — gives each piece of knowledge one authoritative source, while guarding against premature/over-abstraction (Rule of Three). Stack-agnostic.
pr-review	engineering	Slow, adversarial PR/diff review that optimizes for finding real bugs (not approving fast) — built to be run independently by several models and synthesized. Stack- and model-agnostic.
alex-reviewer	engineering	Review frontend code (Vue 3 + TypeScript) the way Alex does — distilled from 1000+ real PR comments. Covers tests, typing, composables & DRY, Vue/i18n idioms, error handling, scope discipline, and review tone.
write-for-human	writing	Write and explain like a human, not an AI — short, answers only what was asked, no flattery, no em-dashes or AI tells, plus plain-words/examples/analogy explanation with a self-audit second pass.
distill	writing	Boil a wall of AI-generated text down to the core — strip the noise, find the one idea, rebuild small with an example and analogy. Answers only what was asked. Stack-agnostic.
grilling	productivity	Interrogate a plan or design one question at a time before building — proposes its own answer to each, checks the codebase first, and drops into learning-worked-example when a question exposes a real knowledge gap.

Installing a skill

Every skill here is a self-contained <skill-name>/SKILL.md folder, and the format is identical across Claude, Cursor, and Gemini — only the install directory differs. Copy (or symlink) the skill folder into the agent's skills directory and it's auto-discovered via its name + description frontmatter.

Agent	Personal (all projects)	Project-scoped
Claude Code	~/.claude/skills/	.claude/skills/
Cursor	~/.cursor/skills/	.cursor/skills/
Gemini CLI	~/.gemini/skills/	.gemini/skills/

name must be lowercase letters/numbers/hyphens and match the folder name; description tells the agent what the skill does and when to use it.

Quick install (one agent)

First clone the repo and cd into it — the commands below symlink/copy out of the working tree, so they must run from inside it:

git clone <repo-url> skills
cd skills

Then symlink (stays in sync on git pull, but needs the clone to stay on disk) or copy (independent snapshot, won't auto-update):

# Claude Code, personal — symlink
ln -s "$PWD/learning-worked-example" ~/.claude/skills/learning-worked-example

# …or copy
cp -r learning-worked-example ~/.claude/skills/

Then start a new session and ask for something the skill's description matches.

Install once, share across all three agents

Cursor and Gemini both read the interoperable ~/.agents/skills/ path, and Cursor also reads ~/.claude/skills/. Put the skill in ~/.agents/skills/ once and symlink it into Claude's directory:

mkdir -p ~/.agents/skills
ln -s "$PWD/learning-worked-example" ~/.agents/skills/learning-worked-example
ln -s ~/.agents/skills/learning-worked-example ~/.claude/skills/learning-worked-example

Now Cursor and Gemini pick it up from ~/.agents/skills/, and Claude from the symlink in ~/.claude/skills/.

Adding a skill

1. Write or pick a research note in the Obsidian skills/ folder.
2. Create <skill-name>/SKILL.md here with name + description frontmatter.
3. Link the note and the skill in both READMEs.