Skills
Reusable agent skills for coding, writing, research, and slide work. This repo is organized for Codex-first workflows and can also be installed as a standard skills repo.
🚀 Quick Start
Install the collection:
npx skills add narumiruna/skills
Then open Codex and run /skills to inspect what was installed. Invoke a skill explicitly with $skill-name, or describe the task normally and let Codex choose a match.
📦 Install
Standard install with npx skills
Use this when you want the collection without linking a local checkout:
npx skills add narumiruna/skills
Standard discovery exposes active skills only; deprecated skills live outside skills/ and remain available solely for repository reference or explicit local use.
🧭 How To Use In Codex
- Run
/skillsto inspect the installed collection. - Type
$managing-python-with-uv,$applying-imrad, or another skill name to invoke one explicitly. - Describe the task normally and let Codex choose a matching skill.
If Codex does not pick up a local skill change, restart Codex and try again.
🧰 Skill Catalog
Python
Repository path: skills/python/
| Skill | Use it for |
|---|---|
managing-python-with-uv |
uv projects, scripts, dependencies, checks, builds, and authorized publishing. |
building-typer-clis |
Typer command grammar, typed parameters, callbacks, packaging, completion, and current tests. |
configuring-python-logging |
Python logging backend, boundary, sink, context, and exception choices. |
using-peewee-orm |
Peewee binding, connection, transaction, and isolated test lifecycles. |
Writing & Research
Repository path: skills/writing-research/
| Skill | Use it for |
|---|---|
writing-plans |
Drafting, executing, and tracking lean plans with acceptance evidence. |
grilling-designs |
Evidence-informed, one-decision-at-a-time design grilling. |
applying-imrad |
Evidence-traceable IMRaD fit checks, reviews, transformations, and drafts. |
creating-telegraph-pages |
Preparing and publishing one explicitly authorized Telegra.ph article. |
grounding-with-google-genai |
Grounded Google Search, Maps, and specific-URL research with Gemini. |
Learning & Explanation
Repository path: skills/learning-explanation/
| Skill | Use it for |
|---|---|
explaining-step-by-step |
Progressive, evidence-grounded mental models for complex material. |
UI/UX Design
Repository path: skills/ui-ux-design/
| Skill | Use it for |
|---|---|
designing-user-interfaces |
Design and review clear, accessible, platform-adapted user interfaces. |
designing-user-experiences |
Plan new and existing digital workflows through an approval-gated UX process. |
iterating-ui-improvements |
Explicitly invoked DevTools audit-fix-commit loops for evidence-backed UI improvements. |
Slides & Visuals
Repository path: skills/slides-visuals/
| Skill | Use it for |
|---|---|
creating-slide-decks |
Complete Marp decks with coordinated narrative, colors, visuals, and rendering. |
authoring-marp-slides |
Focused Marp/Marpit authoring, templates, themes, and rendered checks. |
designing-slide-colors |
Semantic slide palettes with usage rules and measured contrast evidence. |
creating-svg-illustrations |
Accessible, portable SVG diagrams and illustrations for target artifacts. |
creating-mermaid-diagrams |
Editable Mermaid diagrams with optional consumer-ready SVG rendering. |
Workflow & Repository Maintenance
Repository path: skills/workflow-repository/
| Skill | Use it for |
|---|---|
creating-agent-skills |
Creating, naming, renaming, optimizing, and reviewing lean, discoverable agent skills. |
reviewing-code |
Evidence-led read-only review, including edge-case audits and PR preflight, with authorized hardening handoff. |
running-panel-review-loops |
Iterative multi-reviewer code review, verified fixes, and evidence-based acceptance. |
resolving-pr-review-comments |
Explicit PR feedback assessment and local fixes, with exact approval for public actions. |
hardening-code-paths |
Confirming, fixing, and verifying plausible code-path failure modes. |
using-jira-cli |
Read-only Jira inspection and precisely authorized CLI mutations. |
applying-tdd |
Practical red-green-refactor for non-trivial behavior changes. |
writing-git-commits |
Drafting, validating, or creating focused Conventional Commits from diffs. |
syncing-main-branch |
Explicit, state-preserving switch to main and fast-forward-only update. |
managing-git-worktrees |
Loss-aware worktree creation, repair, removal, pruning, and branch cleanup. |
maintaining-memory-md |
Reads and maintains verified, concise repository gotchas and preferences. |
writing-agents-md |
Creating or reviewing lean, scoped, evidence-backed AGENTS.md guidance. |
🗄️ Deprecated Skills
Deprecated skills remain in deprecated/<skill-name>/ for reference and are excluded from standard discovery.
| Skill | Notes |
|---|---|
checking-cli-help |
Legacy decision rule for focused command-help inspection. |
cleaning-atuin-history |
Legacy Atuin audit and exact-approval cleanup preparation. |
building-codex-hooks |
Version-sensitive legacy Codex CLI hook reference. |
writing-work-logs |
Legacy explicit-only Git-evidence work logs. |
naming-agent-skills |
Merged into creating-agent-skills; retained as a compatibility reference. |
researching-gourmet-venues |
Rarely used city dining workflow retained as a compatibility reference. |
No comments yet
Be the first to share your take.