Personal Skills

This repository contains personal Agent Skills compatible with any harness that supports the spec (Claude Code, Pi, and others). Each skill lives under skills/<skill-name>/ and provides a SKILL.md entry point.

Installation

Install this repository with npx skills:

npx skills add https://github.com/rangzen/skills

To update installed skills later:

npx skills update

For local development, you can also install from a checkout:

npx skills add /path/to/skills

Skills

Skill	Description
init-docs	Scaffolds a progressive-disclosure documentation base for a software project (docs/, AGENTS.md).
scoped-commits	Guides commit messages using a scope-first format instead of Conventional Commits.

init-knowledge-directory was merged into the kp-init skill in rangzen/knowledge-project-skills.

init-docs

Scaffolds a progressive-disclosure documentation base for a software project.

Use it when you want to:

• initialize docs/ for a repo
• create AGENTS.md and architecture/design/product docs
• set up a documentation structure without overwriting existing files

Typical prompts:

• /init-docs
• set up docs for this project
• initialize the knowledge base for this repo

scoped-commits

Guides commit messages using a scope-first format instead of Conventional Commits.

Format:

scope: short description

Use it when you want help:

• writing a commit message
• defining commit conventions
• reviewing whether a commit title is well scoped

Typical prompts:

• write a commit message for these changes
• what commit style should we use?

External Skills

Omit --skill <name> to open an interactive multi-select picker for the whole repo:

npx skills add anthropics/skills

Skills from anthropics/skills worth installing:

Skill	Description	Install
skill-creator	Create, modify, and measure agent skills.	npx skills add anthropics/skills --skill skill-creator
pdf	Read, merge, split, fill, encrypt, and OCR PDF files.	npx skills add anthropics/skills --skill pdf
docx	Create, read, and edit Word documents.	npx skills add anthropics/skills --skill docx
xlsx	Open, read, edit, and create spreadsheets (.xlsx, .csv, .tsv).	npx skills add anthropics/skills --skill xlsx
pptx	Work with PowerPoint files as input or output.	npx skills add anthropics/skills --skill pptx

Skills from rangzen/knowledge-project-skills:

Skill	Description	Install
kp-init	Scaffold a new knowledge project: directory structure, .knowledge-project config, and stub files.	npx skills add rangzen/knowledge-project-skills --skill kp-init
kp-source	Add and track sources (PDF, URL, CSV, database) with ingestion dates and hashes.	npx skills add rangzen/knowledge-project-skills --skill kp-source
kp-staging	Run LLM extractors over ingested sources and write structured JSON to staging/.	npx skills add rangzen/knowledge-project-skills --skill kp-staging
kp-query	Query the knowledge base and save answers with provenance to wiki/queries/.	npx skills add rangzen/knowledge-project-skills --skill kp-query
kp-wiki	Build the navigable wiki: entity pages, glossary, wikilinks, and typed entry points.	npx skills add rangzen/knowledge-project-skills --skill kp-wiki

Skills from kunchenguid/lavish-axi by Kun Chen:

Skill	Description	Install
lavish	Turn complex or visual agent responses into rich, reviewable HTML artifacts the user can annotate and send feedback on.	npx skills add kunchenguid/lavish-axi --skill lavish

Context

The context/ directory contains personal context files that agents can load to understand who they are working with and how to act on behalf of Cédric.

File	Purpose
context/AGENTS.md	General agent instructions following the agents.md standard. Loaded by compatible agents automatically.
context/OPINIONS.md	Cédric's engineering and product opinions. Agents read this when making technical decisions that benefit from his viewpoint.
context/VOICES.md	Cédric's writing voice profile. Agents read this when writing or posting on his behalf.

Run scripts/link-context.sh to symlink all context files into ~ and create ~/CLAUDE.md -> ~/AGENTS.md.

Inspired by Kun Chen's approach to agent ergonomics and personal context design.

Usage

These skills are trigger-based. After installation, invoke them by naming the skill directly or by asking for the underlying task in natural language.

Examples:

/init-docs
initialize a knowledge directory in this folder
write a scoped commit message for the current diff