Add skill: agent-workflow-packager

[MemoAsh]

Add Skill: agent-workflow-packager

Checklist

Before submitting, ensure your skill meets these requirements:

Frontmatter (SKILL.md YAML header)

• name ? Lowercase kebab-case only (a-z, 0-9, -), max 64 chars
• description ? States WHAT it does and WHEN to use it, includes trigger words, max 1024 chars, third person (no "I"/"you"/"we")
• license ? Present (e.g., Apache-2.0)
• compatibility ? Concrete system requirements or "No special requirements"
• metadata.author ? Your GitHub username
• metadata.version ? Semantic version ("1.0.0")
• metadata.category ? One of the supported categories
• metadata.tags ? Array of 3-5 relevant tags

Body Sections

• Overview ? 2-3 sentence summary
• Instructions ? Step-by-step actionable workflow
• Examples ? At least 2 concrete input/output examples with realistic data
• Guidelines ? Best practices, edge cases, error handling

Quality

• Under 300 lines total
• Instructions are specific enough that an AI agent can follow them without guessing
• Works cross-platform
• Solves a real, recurring problem a developer would encounter at least weekly

Files

• skills/agent-workflow-packager/SKILL.md added
• Use-case article added in use-cases/

---

Which use case does this skill serve?

This serves teams that repeatedly paste the same high-context coding-agent prompts or AGENTS.md sections for PR reviews, releases, CI triage, or repo-specific workflows. The skill turns those repeated workflows into reusable SKILL.md packages with triggers, boundaries, examples, and verification checks.

Added use case: use-cases/package-repeated-ai-agent-workflows.md.

Example usage

A team has a recurring payment PR review prompt covering Stripe webhook replay, idempotency, auth checks, and billing tests. They ask the agent to package that prompt plus old review comments into skills/payment-pr-reviewer/SKILL.md. The generated skill includes concrete triggers, required inputs, review steps, approval boundaries, two examples, and a done checklist.

Category

development

Additional context

I also updated the README skill table and use-case list. I ran local checks for frontmatter presence, required fields, target link paths, line counts, non-ASCII content, and git diff --check. I did not include a regenerated skills/index.json because scripts/generate-index.js appears to parse only LF frontmatter on this Windows checkout and incorrectly reduced the index to one skill.

[Andrew1326]

Reviewed via skill-reviewer (ACCEPT, quality 93/100, security SAFE, no duplicates) and validator (impact 9.00x, gpt-oss:20b-64k). Frontmatter, structure, examples, and cross-platform checks all pass. Merging with generated _scores.json and regenerated index.json.