karsonenns

Skill Framework — DevOps skill for Claude Code

DevOps community

Organize, validate, and deploy AI agent skills (SKILL.md) at scale — lint rules, token budgets, secrets governance, and one-tree deploys to Claude Code, Codex, Gemini CLI, and Cursor.

How to install Skill Framework

This entry records only its repository, not the path inside it, so there is no exact command to give. Open karsonenns/skill-framework and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Skill Framework does

Organize, validate, and deploy AI agent skills (SKILL.md) at scale — lint rules, token budgets, secrets governance, and one-tree deploys to Claude Code, Codex, Gemini CLI, and Cursor.

Alternatives in DevOps

  • Skypilot — Run, manage, and scale AI workloads on any AI infrastructure 9.7k ★
  • Claudable — Claudable is an open-source web builder that leverages local CLI agents, such as Claude Code, Codex, Gemini CL 4k ★
  • Azure Batch Java — Large-scale parallel and HPC batch jobs 1.8k ★

README

Skill Framework — organize, validate, and deploy AI agent skills (SKILL.md) at scale

[![CI](https://img.shields.io/github/actions/workflow/status/karsonenns/skill-framework/ci.yml?style=flat-square&label=CI&labelColor=1c2128)](https://github.com/karsonenns/skill-framework/actions/workflows/ci.yml) [![npm](https://img.shields.io/npm/v/skillfw?style=flat-square&labelColor=1c2128&color=0b8a68)](https://www.npmjs.com/package/skillfw) ![node >= 20](https://img.shields.io/badge/node-%E2%89%A5%2020-0b8a68?style=flat-square&labelColor=1c2128) ![runtime deps: 1](https://img.shields.io/badge/runtime_deps-1-0b8a68?style=flat-square&labelColor=1c2128) [![License: MIT](https://img.shields.io/badge/license-MIT-0b8a68?style=flat-square&labelColor=1c2128)](LICENSE)

**One source tree of skills, compiled to every runtime that reads `SKILL.md`.**

The [Agent Skills spec](https://agentskills.io) defines a single skill folder — and stops there. Nothing defines how hundreds of skills compose into an organization, so teams get duplicated context, token bloat, committed credentials, and no idea whether the skills an agent is running match the repo. Skill Framework is that missing layer: a **convention**, a **linter** that enforces it in CI, and a **deploy engine** that compiles one source tree to every runtime that reads `SKILL.md`. Fully offline; no LLM calls.

npx skillfw init && sf lint && sf deploy --dry-run
Terminal session: npx skillfw init scaffolds a tree, sf lint passes 3 skills with no problems, sf deploy --dry-run prints a Terraform-style plan of 16 files to add across 4 targets

The convention

Skills are class