AGENTS.md

This repository is a collection of reusable agent skills following the Agent Skills open standard.

Repository structure

agent_skills/
├── AGENTS.md                          # This file — agent index
├── README.md                          # Human-readable overview
├── package.json                       # Shared Node.js dependencies for all skills
├── .env                               # Credentials (git-ignored, not readable by Claude)
├── .env.example                       # Credential template
├── .gitignore                         # Ignores: outputs/, reports/, node_modules/, .env
├── .claudeignore                      # Prevents Claude from reading .env
├── docs/                              # Detailed reference documentation
│   ├── setup.md                       # Installation and credential setup
│   ├── download-website-to-markdown.md
│   └── confluence-operations.md
├── outputs/                           # Downloaded pages and generated content (git-ignored)
│   └── reference-docs/               # Fetched reference documentation
├── reports/                           # Markdown files uploaded to Confluence (git-ignored)
├── download-website-to-markdown/
│   ├── SKILL.md                       # Skill definition (frontmatter + agent instructions)
│   └── scripts/
│       └── download.ts               # Playwright-based downloader
├── confluence-operations/
│   ├── SKILL.md                       # Skill definition (frontmatter + agent instructions)
│   └── scripts/
│       ├── upload.ts                 # Confluence REST API uploader
│       └── download.ts              # Confluence REST API downloader (page → Markdown)
├── sales-agent/
│   └── SKILL.md                       # Skill definition (frontmatter + agent instructions)
├── pdf-to-markdown/
│   ├── SKILL.md                       # Skill definition (frontmatter + agent instructions)
│   └── scripts/
│       └── convert.ts                # pdftotext-based PDF converter
├── english-tech-blog-structure/
│   ├── SKILL.md                       # Skill definition (frontmatter + agent instructions)
│   └── NOTICE                         # Apache-2.0 attribution (getsentry/skills)
├── llm-evaluation/
│   └── SKILL.md                       # Skill definition (frontmatter + agent instructions)
├── linkedin-article/
│   └── SKILL.md                       # Skill definition (frontmatter + agent instructions)
├── devto-article/
│   └── SKILL.md                       # Skill definition (frontmatter + agent instructions)
├── zenn-article/
│   └── SKILL.md                       # Skill definition (frontmatter + agent instructions)
└── md-deck/
    └── SKILL.md                       # Skill definition (frontmatter + agent instructions)

Available skills

download-website-to-markdown

Downloads a URL to a local Markdown file using a headless Chromium browser. Full reference: docs/download-website-to-markdown.md

cd agent_skills
npx tsx download-website-to-markdown/scripts/download.ts <url> [output]

confluence-operations

Upload or download Confluence pages as local Markdown files. Full reference: docs/confluence-operations.md

cd agent_skills
npx tsx confluence-operations/scripts/upload.ts <file.md> [--title "..."] [--parent-id <id>] [--page-id <id>]
npx tsx confluence-operations/scripts/download.ts <page-id> [output]

sales-agent

Sales intelligence skill — query deals, pipeline, rep performance, and conversation transcripts via Cortex Agent and Cortex Analyst. Uses built-in Cortex Code tools only (no custom scripts required).

pdf-to-markdown

Converts a PDF file to Markdown using pdftotext (poppler). Outputs clean Markdown with YAML front matter and heading detection heuristics. Requires brew install poppler.

cd agent_skills
npx tsx pdf-to-markdown/scripts/convert.ts <input.pdf> [output.md]

english-tech-blog-structure

Structure guide for English technical blog posts (openings, headings, technical quality, titles, closings). Pairs with blader/humanizer for sentence-level AI tells. Adapted from getsentry/skills blog-writing-guide (Apache-2.0); see english-tech-blog-structure/NOTICE.

llm-evaluation

Report LLM and pipeline evaluation numbers honestly in technical articles: sample size, exact conditions, and no invented figures. Read it when a draft cites accuracy, latency, token, or cost measurements. Pairs with english-tech-blog-structure, linkedin-article, and devto-article.

linkedin-article

Draft and revise English LinkedIn technical articles (Pulse) and feed posts. Pairs with english-tech-blog-structure and blader/humanizer.

devto-article

Write, revise, and draft-register English technical articles for dev.to. Pairs with english-tech-blog-structure and blader/humanizer.

zenn-article

Plan, write, and revise Japanese technical articles for Zenn, including html-embed HTML/CSS diagrams. Instructions: zenn-article/SKILL.md.

md-deck

Create and revise 16:9 PDF slide decks with md-deck: workflow, style versus content, spacing density, and review before tag or release. Palette: visual-color. Diagram principles: visual-diagram. Instructions: md-deck/SKILL.md.

visual-color

Shared dark-blue palette for md-deck themes, article covers, inline diagrams, and generated images. Instructions: visual-color/SKILL.md.

visual-diagram

Reading order, short labels, and a text equivalent for process and structure figures on slides and in articles. Instructions: visual-diagram/SKILL.md.

Setup

See docs/setup.md for full installation and credential setup.

Quick setup:

cd agent_skills
npm install
npx playwright install chromium
cp .env.example .env   # then fill in Confluence credentials

Creating or modifying skills

Follow the Agent Skills open specification. Key rules:

• Each skill lives in its own directory. The directory name must exactly match the name field in SKILL.md.
• SKILL.md must have YAML frontmatter with at minimum name and description, followed by Markdown instructions.
• name: lowercase letters, numbers, and hyphens only; no consecutive hyphens; max 64 characters.
• description: describes what the skill does and when to use it; max 1024 characters.
• Optional frontmatter fields: license, compatibility, metadata, allowed-tools.

Validating a skill

Use npx skills-ref validate to check a skill against the spec before committing:

cd agent-skills
npx skills-ref validate ./<skill-name>

Validate all skills at once:

for d in */; do npx skills-ref validate "./${d%/}" 2>&1; done