om-create-agents-md

Creates and refactors AGENTS.md agent guideline files for Open Mercato packages and modules.

1.7k|382|Updated Sep 10, 2025
One-click install
npx skills add https://github.com/open-mercato/open-mercato --skill om-create-agents-md
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: om-create-agents-md
Source: https://github.com/open-mercato/open-mercato/tree/main/.ai/skills/om-create-agents-md
Command: npx skills add https://github.com/open-mercato/open-mercato --skill om-create-agents-md

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Writing AGENTS.md files for coding agents often produces descriptive documentation instead of actionable instructions, leading to inconsistent agent behavior across packages and modules. This Skill enforces a single prescriptive convention so every AGENTS.md tells agents exactly what to do, what to ask about, what never to do, and how to validate their work.

Core Features & Use Cases

  • Boundary Convention Enforcement: Guarantees every AGENTS.md contains the four mandatory sections — Always, Ask First, Never, and Validation Commands — in the correct order with size-scaled minimum rule counts.
  • Prescriptive Tone Transformation: Converts descriptive prose ("The module provides...") into imperative MUST/MUST NOT rules, decision tables, and numbered checklists.
  • Legacy Migration: Migrates pre-existing AGENTS.md files with legacy headings (MUST Rules, Critical Rules) into the boundary convention without losing high-risk operational details.
  • Use Case: When adding a new Open Mercato package or module, invoke this Skill to generate a compliant AGENTS.md with correct structure, tone, validation commands, and a root Task Router update.

Quick Start

Ask the agent to create an AGENTS.md for your new Open Mercato package following the boundary convention with Always, Ask First, Never, and Validation Commands sections.

Frequently Asked Questions about om-create-agents-md

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write an AGENTS.md file for a coding agent?

Structure it with four mandatory sections: Always (MUST/MUST NOT rules), Ask First (decisions needing maintainer approval), Never (prohibited actions), and Validation Commands (runnable test/build commands). Use imperative verbs and numbered checklists instead of descriptive prose.

What sections must every AGENTS.md file contain?

Every AGENTS.md must contain exactly one of each: ## Always, ## Ask First, ## Never, and ## Validation Commands, in that order. Small files need at least 3 Always rules, 1 Ask First item, 2 Never items, and 1 validation command.

How do I migrate a legacy AGENTS.md to the boundary convention?

Hoist existing rules into the correct boundary: MUST/MUST NOT defaults go to Always, scope-changing decisions become Ask before items, and hard prohibitions become Never bullets. Add a Validation Commands block and verify headings with grep.

What tone should agent guideline files use?

Use prescriptive, imperative tone: start sections with verbs like Use, Add, Configure, or constraints like MUST and MUST NOT. Never start with descriptive openers like "The module provides" — every sentence should tell the agent what to do.

When should rules go in Ask First versus Never?

Ask First holds decisions needing maintainer approval, such as changing dependencies, contracts, or defaults. Never holds hard prohibitions like skipping tenant scoping or editing generated files. Prefer the most restrictive boundary when a rule fits both.