How do I install the Instruction Guide skill?

Run npx skillmds@latest add xonovex/instruction-guide in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Instruction Guide safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Instruction Guide?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Instruction Guide free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Instruction Guide?

xonovex (@xonovex) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Instruction Guide by xonovex — Agent Skill

Name: Instruction Guide
Author: xonovex

Instruction Guide

Use when authoring, reviewing, initializing, syncing, simplifying, consolidating, or assimilating AGENTS.md project-instruction files. Triggers on edits to `AGENTS.md` and on prompts about bootstrapping fresh instructions, refreshing stale docs to match the filesystem, condensing verbose files, deduping across the repo, or porting organizational patterns from another project, even when the user doesn't say 'instructions' or 'AGENTS.md'.

xonovex 9bebf98 9 files · 22.8 KB Updated Sep 17, 2026

File contents

Project Instruction Guidelines

Author, refresh, simplify, consolidate, and port AGENTS.md project-instruction files while preserving technology names, paths, and project context.

Core Principles

Open standard: AGENTS.md is the cross-harness format read by many agent harnesses; keep content harness-neutral
README for agents: operational detail (commands, conventions, gotchas) belongs in AGENTS.md; human-onboarding prose belongs in README.md
Teammate heuristic: "anything you'd tell a new teammate" belongs here; anything you'd tell a casual visitor belongs in README
Nested precedence: for monorepos, the AGENTS.md closest to the edited file wins; place subproject-specific guidance in subdirectories
Commands are executable: agents auto-run programmatic checks listed in AGENTS.md; treat command examples as instructions, not just docs
Advisory scope: AGENTS.md communicates project guidance; it does not itself execute automation
Preserve Project Context: never modify technology names, paths, or commands
Match Style and Voice: maintain the target file's formatting and terminology
Structure Integrity: keep section order and hierarchy intact; preview changes (dry-run) before applying
Signal Over Volume: keep only non-obvious content; anything inferable from package.json + source belongs in the code, not the doc
Deletion-first editing: a simplify or consolidate pass cuts at least as much as it adds, and reuses existing vocabulary rather than coining new terms; skill-guide owns this rule
Living documentation: update alongside project changes (sync operation)

Gotchas

AGENTS.md files that restate code are worse than no doc. They rot when the code changes
Tech names look like noise but are essential; never substitute generics for moon / Terraform / npm
Root AGENTS.md is the project entry-point doc, never auto-delete even if it looks thin
Manual descriptions in Subdirectories often encode info the filesystem can't (purpose, owner, status): preserve verbatim during sync
Nested AGENTS.md aren't duplicates. They're scope-specific overrides; never merge a subproject's file into the root
Build / test commands listed in AGENTS.md will be executed by agents; broken commands waste agent turns
"Always do X" in AGENTS.md enforces nothing. The model may not comply and may never load the file; capitalizing MUST adds no executor, and the failure is silent because the rule never ran

Progressive Disclosure

Read references/init.md - Load when bootstrapping a fresh AGENTS.md for a directory
Read references/sync.md - Load when refreshing AGENTS.md to match current filesystem state
Read references/simplify.md - Load when condensing a verbose AGENTS.md
Read references/consolidate.md - Load when cleaning up scattered AGENTS.md across a repo
Read references/merge.md - Load when porting organizational patterns from another project's AGENTS.md

---name: instruction-guidedescription: Use when authoring, reviewing, initializing, syncing, simplifying, consolidating, or assimilating AGENTS.md project-instruction files. Triggers on edits to `AGENTS.md` and on prompts about bootstrapping fresh instructions, refreshing stale docs to match the filesystem, condensing verbose files, deduping across the repo, or porting organizational patterns from another project, even when the user doesn't say 'instructions' or 'AGENTS.md'.---# Project Instruction GuidelinesAuthor, refresh, simplify, consolidate, and port AGENTS.md project-instruction files while preserving technology names, paths, and project context.## Core Principles- **Open standard**: AGENTS.md is the cross-harness format read by many agent harnesses; keep content harness-neutral- **README for agents**: operational detail (commands, conventions, gotchas) belongs in AGENTS.md; human-onboarding prose belongs in README.md- **Teammate heuristic**: "anything you'd tell a new teammate" belongs here; anything you'd tell a casual visitor belongs in README- **Nested precedence**: for monorepos, the AGENTS.md closest to the edited file wins; place subproject-specific guidance in subdirectories- **Commands are executable**: agents auto-run programmatic checks listed in AGENTS.md; treat command examples as instructions, not just docs- **Advisory scope**: AGENTS.md communicates project guidance; it does not itself execute automation- **Preserve Project Context**: never modify technology names, paths, or commands- **Match Style and Voice**: maintain the target file's formatting and terminology- **Structure Integrity**: keep section order and hierarchy intact; preview changes (dry-run) before applying- **Signal Over Volume**: keep only non-obvious content; anything inferable from `package.json` + source belongs in the code, not the doc- **Deletion-first editing**: a simplify or consolidate pass cuts at least as much as it adds, and reuses existing vocabulary rather than coining new terms; **skill-guide** owns this rule- **Living documentation**: update alongside project changes (sync operation)## Gotchas- AGENTS.md files that restate code are worse than no doc. They rot when the code changes- Tech names look like noise but are essential; never substitute generics for `moon` / `Terraform` / `npm`- Root AGENTS.md is the project entry-point doc, never auto-delete even if it looks thin- Manual descriptions in Subdirectories often encode info the filesystem can't (purpose, owner, status): preserve verbatim during sync- Nested AGENTS.md aren't duplicates. They're scope-specific overrides; never merge a subproject's file into the root- Build / test commands listed in AGENTS.md will be **executed** by agents; broken commands waste agent turns- "Always do X" in AGENTS.md enforces nothing. The model may not comply and may never load the file; capitalizing MUST adds no executor, and the failure is silent because the rule never ran## Progressive Disclosure- Read [references/init.md](references/init.md) - Load when bootstrapping a fresh AGENTS.md for a directory- Read [references/sync.md](references/sync.md) - Load when refreshing AGENTS.md to match current filesystem state- Read [references/simplify.md](references/simplify.md) - Load when condensing a verbose AGENTS.md- Read [references/consolidate.md](references/consolidate.md) - Load when cleaning up scattered AGENTS.md across a repo- Read [references/merge.md](references/merge.md) - Load when porting organizational patterns from another project's AGENTS.md

Frequently asked questions