Skip to content

Introduction

agent-config is a content + governance layer for AI coding agents — a curated library of skills, rules, commands, guidelines and personas, distributed on npm as @event4u/agent-config and projected into each host tool’s native config format. It follows the Agent Skills open standard and compiles into 7+ host agents (Claude Code, Cursor, Augment, Cline, Windsurf, Gemini CLI, GitHub Copilot, and more).

Think of it as a portable playbook and style guide those tools read — not a replacement for them. You bring your own AI provider; agent-config brings the depth and the discipline.

Artifact Count What it is
Skills 278 On-demand expertise and executable procedures, matched to a task by intent
Rules 105 Always-active behaviour constraints auto-loaded every conversation
Commands 190 Slash-command workflow orchestrators (/commit, /create-pr, /work, …)
Guidelines 101 Reference conventions read on demand
Personas 29 Review lenses dispatched during code/design review
Advisors 5 Specialist advisory lenses

A capability router loads the right skill on intent rather than dumping the whole library into context. See Architecture for the full mental model.

A content + governance layer, not a runtime:

  • No background daemon — the host tool runs the agent loop.
  • No separate state database and no self-rewriting memory — every learned change is human-reviewed.
  • No auto-build pipeline — the same layer stays portable across tools.

“Zero runtime daemon” is itself a machine-checked claim. Every public capability or number binds to resolvable evidence — verify it yourself.

The skills, rules and personas are distilled from real production work on TypeScript and PHP codebases. The governance mechanics are stack-agnostic, but the domain heuristics are richest where they were forged — treat coverage on other stacks as promising, not proven.