Generic, reusable agent skills for Claude Code. Each skill is a self-contained workflow — a SKILL.md with a name, trigger description, and step-by-step instructions.
Skills are designed to be portable. They contain no personal configuration, hardcoded paths, or user-specific details. Anything personal lives in a separate private workspace alongside these skills.
| Skill | Description |
|---|---|
code-implement |
TDD implementation from a plan — worktree workflow, write tests first, implement, review |
code-init |
Codebase orientation — read docs, map structure, summarize architecture |
code-plan |
Plan a code change via structured Q&A — clarify, approach, write plan |
code-review |
Review code for plan adherence, design, and clean code |
delegate |
Delegate a task to a background coding agent via acpx — second opinions, background work, parallel exploration |
download-video |
Download a video to Desktop via yt-dlp |
find-skills |
Search and install skills from the skills.sh ecosystem |
go |
Quick launcher — jump to a file, URL, or bookmark by shortcut name |
learn |
Persist new knowledge — facts, prefs, behaviors, skills, system changes |
log |
Save a detailed session log |
read-later |
Save a URL as a clean readable PDF |
recall |
Search workspace for past context from previous sessions |
review |
GTD-style daily/weekly review — tasks, calendar, plans, priorities |
Skills work best as part of a broader agent workspace. Here's the setup used by this author:
workspace/
├── AGENTS.md # System prompt — identity, routing, preferences
├── docs/ # Personal config (filing rules, calendar IDs, contacts, etc.)
├── logs/ # Session logs (YYYY-MM-DD Topic.md)
├── skills/ # ← this repo (symlinked or cloned here)
├── jobs/ # Background job definitions
└── user/ # Local content (notes vault, etc.) — not synced
Three layers:
AGENTS.md— loaded every session. Identity, routing, high-level preferences. Keeps short; points to docs for detail.docs/— personal config read on demand. Calendar IDs, contact shortcuts, filing rules, tool-specific settings. Skills reference this with "refer to any provided documentation."skills/— this repo. Generic workflows that work for anyone. No personal config baked in.
This separation means skills can be public while docs and AGENTS.md stay private.
Syncing: The workspace (minus user/) syncs across machines via Stash — a CRDT-based file sync tool backed by GitHub. Skills live in their own repo and are symlinked into the workspace.
Each skill is a directory with a SKILL.md:
skills/
└── my-skill/
└── SKILL.md
SKILL.md structure:
---
name: my-skill
description: One sentence — when should this skill trigger?
---
What this skill does, in one line.
## Steps
1. Step one
2. Step two
...Keep skills generic. If a skill needs user-specific config (paths, IDs, contacts), document that the skill expects it to be provided in context — don't hardcode it.