agent-skills-playbook

agent-skills-playbook

Production-grade AI agent skills, prompts, and operating playbooks.

This repo is inspired by the late-April 2026 wave around agent skill libraries and coding-agent behavior packs, including GitTrend's April 24 ranking for forrestchang/andrej-karpathy-skills, google/skills, and addyosmani/agent-skills.

Goal

Build a practical catalog of small, reusable agent skills that make coding agents more reliable in real work: research, planning, implementation, review, release notes, documentation, and incident debugging.

First Skills

• research-first-dev: gather sources and repo context before implementation.
• code-review-hawk: review changes with line-level findings and risk ranking.
• ship-ready-readme: turn rough project notes into a usable README.
• agent-memory-brief: create compact handoff notes for long-running agent tasks.
• security-pass: scan a repo for obvious secret, dependency, and permission risks.

Repository Shape

skills/
  research-first-dev/
    SKILL.md
  code-review-hawk/
    SKILL.md
templates/
  handoff.md
  release-notes.md
examples/
  sample-agent-run.md

Milestone 1

• Done: define the SKILL.md template.
• Done: add five starter skills.
• Done: include a validation CLI.
• Next: add before/after examples for each skill.

Run It

python -m agent_skills_playbook.cli list
python -m agent_skills_playbook.cli validate
python -m agent_skills_playbook.cli render-index

Test It

PYTHONPATH=src python -m pytest

Publish

This package is ready for GitLab Package Registry and PyPI releases. See RELEASE.md.

Trend Notes

• Agent skills are getting shared as portable behavior modules.
• The viral repos are simple, text-first, and easy to fork.
• The opportunity here is quality control: testable, versioned skills instead of random prompt snippets.