Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Sign up
Appearance settings

dimohy/AUTHORING.DIMOHY

Folders and files

NameName
Last commit message
Last commit date

Latest commit

History

8 Commits

Repository files navigation

AUTHORING.DIMOHY

AUTHORING.DIMOHY is a meta authoring guide for building consistent AI-agent customization workflows for any project. It helps agents create, update, and audit AGENTS.md, SPECS.md, rules/instructions, prompts(commands), chat modes, skills, hooks, subagents, and MCP/tool integrations.

Canonical repository: https://github.com/dimohy/AUTHORING.DIMOHY

Korean guide: ko/README.md · Korean source directory: ko/

AUTHORING.DIMOHY workflow overview

Guide locations

Language Location
English en/ — use the newest AUTHORING.DIMOHY.r*.md file
Korean ko/ — use the newest AUTHORING.DIMOHY.r*.md file

Last updated: 2026年04月28日

When should agents read this guide?

Do not keep the full AUTHORING guide loaded during ordinary coding or writing sessions. Read it only when you are doing customization work, such as:

  • bootstrapping AGENTS.md, SPECS.md, or .github/ customization files for a new project;
  • realigning existing rules, prompts(commands), skills, agents, hooks, and MCP/tool settings;
  • auditing customization file structure, category coverage, or workflow consistency;
  • updating AUTHORING.DIMOHY itself or bumping its revision.

During normal project work, the lightweight AGENTS.md in that project is the primary operating constitution.

Quick start

1. Install into a new project

  1. To use AUTHORING, open the canonical repository on the web, choose the newest localized guide from en/ or ko/, then copy/download it into the target project root or reference that web file explicitly.
    • English: newest en/AUTHORING.DIMOHY.r*.md
    • Korean: newest ko/AUTHORING.DIMOHY.r*.md
  2. In the target project, you may optionally keep the copied file as AUTHORING.DIMOHY.md for a stable local reference.
  3. Ask your AI agent to bootstrap the project.
    • Read AUTHORING.DIMOHY.md and create AGENTS.md for this project
  4. The agent should create AGENTS.md, optionally create SPECS.md, and offer project-appropriate agents, prompts, skills, hooks, chat modes, and tool settings.

2. Run interactive bootstrap

Use this when you want the agent to interview you before generating files:

  • Run AUTHORING

Interactive bootstrap runs four steps:

  1. clarify project purpose, audience, and outputs;
  2. propose the recommended structure;
  3. interview for SPECS.md details;
  4. generate approved files after confirmation.

3. Realign an existing project

Use this when a project already has AGENTS.md, .github/, .claude/, .cursor/, or similar customization files:

  • realign this project with AUTHORING

The agent should classify each asset as keep / update / remove / missing, present a change proposal, apply only approved changes, and re-audit afterward.

4. Suggest useful improvements upstream

If using AUTHORING in another project leads to a generally useful improvement, the upstream path is an Issue. External users submit reusable suggestions through Issues.

Ask the agent in the target project:

  • this AUTHORING improvement is useful upstream
  • create an AUTHORING improvement issue

The agent should let the user choose:

  1. Use only / no upstream — keep the improvement in the current project only.
  2. Update local AUTHORING copy — edit the copied AUTHORING.DIMOHY*.md in the current project only.
  3. Create upstream Issue — create an Issue in https://github.com/dimohy/AUTHORING.DIMOHY when authenticated GitHub tooling is available.
  4. Prepare Issue draft — generate the Issue title/body for manual submission when authentication or browser automation is unavailable.
  5. Skip / later — do nothing now.

This repository includes .github/ISSUE_TEMPLATE/authoring-improvement.yml for structured Issue submissions. Maintainers can then use .github/prompts/triage-authoring-issues.prompt.md to read registered Issues, judge whether they are generally useful, and apply accepted changes to AUTHORING.DIMOHY.

Managed categories

Category Example paths Purpose
Operating constitution AGENTS.md Session, response style, ask_user, harness rules
Domain specs SPECS.md Project-specific outputs, paths, commands, acceptance criteria
Rules / Instructions .github/instructions/, .cursor/rules/, .claude/rules/ Automatically injected context rules
Prompts / Commands .github/prompts/, .claude/commands/ Reusable slash commands or task templates
Chat Modes .github/chatmodes/ Special-purpose chat modes
Skills .github/skills/<name>/SKILL.md, .claude/skills/<name>/SKILL.md, .agents/skills/<name>/SKILL.md Procedural workflows with bundled assets
Hooks .github/hooks/, .claude/hooks/ Lifecycle checks or transformations
Agents .github/agents/, .claude/agents/ Specialized subagents
MCP / Tools .vscode/mcp.json, .mcp.json External docs, browser, APIs, build/test tools

TaskSync session-number fix

The current guide includes explicit TaskSync safeguards and the Issue-only upstream contribution flow:

  • reuse the injected or returned session_id;
  • do not resend session_id: "auto" during a normal conversation;
  • do not confuse UI labels such as AGENT 1 or AGENT 2 with session_id;
  • if the ask_user schema has no options field, write numbered choices in the question body;
  • preserve the exact session_id during summaries, handoffs, and compaction;
  • when a reusable AUTHORING improvement is found, let the user choose local-only use, local AUTHORING copy update, upstream Issue creation, Issue draft preparation, or skip/later.

Revision and localization policy

AUTHORING.DIMOHY uses simple revision identifiers: r1, r2, r3, ... . The r prefix is part of the identifier.

Whenever a revision changes, update all of the following together:

  • en/AUTHORING.DIMOHY.rNN.md
  • ko/AUTHORING.DIMOHY.rNN.md
  • this README.md
  • ko/README.md
  • Git tag rNN
  • remote main branch

A release is not complete if only one language is updated.

Revision history is recorded in HISTORY.md. Korean history is recorded in ko/HISTORY.md.

To reduce revision-churn, user-facing docs should avoid hard-coding the current revision number. Outside the localized AUTHORING file title/metadata, the revisioned filename itself, and revision history entries, prefer language directory links such as en/ and ko/, or generic selectors such as AUTHORING.DIMOHY.r*.md.

License

Apache License 2.0. See LICENSE for the full text.

About

A meta-authoring guide for creating consistent, safe, and reusable AI agent instructions.

Resources

License

Stars

Watchers

Forks

Packages

Contributors

AltStyle によって変換されたページ (->オリジナル) /