agents-md

Name: agents-md
Availability: InStock
Author: bpawlakj

Generate anchored AGENTS.md files with optional CLAUDE.md and Copilot shims.

3|Updated Mar 31, 2026

Search Tags:#Claude Code #anti-duplication #agents-md #agent rules #prompt governance #Copilot shim #anchored instructions

Ready to run

npx skills add https://github.com/bpawlakj/ai-devkit --skill agents-md-bpawlakj

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agents-md
Source: https://github.com/bpawlakj/ai-devkit/tree/main/claude/skills/agents-md
Command: npx skills add https://github.com/bpawlakj/ai-devkit --skill agents-md-bpawlakj

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill prevents agents from drifting by generating a single, project-specific AGENTS.md/CLAUDE.md instruction layer that is only as broad as what the repository itself can justify.

Core Features & Use Cases

Anchored, non-duplicative rule authoring: proposes agent rules only when they pass the “test of inclusion” against ai-devkit’s auto-active language rules and the project’s own docs/incidents/specs.
Canonical rules + import shims: writes AGENTS.md as the source of truth, then optionally creates a CLAUDE.md import shim and a Copilot instructions shim that points back to AGENTS.md.
Interactive, safety-first workflow: verifies whether docs exist, inventories existing auto-active rules, scans for anchorable content, and asks before appending/auditing/overwriting.

Quick Start

Ask the agents-md skill to initialize project agent rules by targeting your repo root or a specific scope path.

Frequently Asked Questions about agents-md

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

FAQPage Schema

How do I create anchored project agent rules for Claude Code without duplicating global instructions?▼

To create anchored project agent rules without duplicating global instructions, generate an AGENTS.md file validated by a test-of-inclusion gate against auto-active ~/.claude/rules, ensuring rules match project-specific constraints extracted from specs and architecture docs.

What is an AGENTS.md file and when do I need it for AI agent governance?▼

An AGENTS.md file is a single source of truth for project-specific agent constraints, needed during repo onboarding or when enforcing domain business rules across critical decision points where the agent cannot reliably infer behavior from code alone.

Can I use a Copilot instructions shim with an AGENTS.md rules file?▼

Yes, you can use a Copilot instructions shim alongside an AGENTS.md file. The workflow writes AGENTS.md as the canonical source and optionally generates a Copilot instructions shim that points back to it for consistent behavior.

How do I generate a CLAUDE.md import shim from existing project specifications?▼

You generate a CLAUDE.md import shim by inventorying existing auto-active rule layers, extracting anchorable content from product specs or architecture docs, and writing a shim that imports the canonical AGENTS.md file.

What are the limitations of relying on auto-active language rules for agent behavior?▼

Relying solely on auto-active language rules risks agent drift because it lacks project-specific context. Limitations include missing local conventions and domain business rules, which requires appending an anchored instruction layer to enforce deterministic authorship.