# Dongjiyong Workstyle

> ---

- Skill: `dongjiyong/dongjiyong-workstyle` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add dongjiyong/dongjiyong-workstyle`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dongjiyong/dongjiyong-workstyle/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: dongjiyong (https://skillmd.com/u/dongjiyong)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/dongjiyong/dongjiyong-workstyle

---

---
name: dongjiyong-workstyle
description: Apply Dongjiyong's preferred collaboration style by default for general coding, documentation, design, debugging, planning, and research tasks unless stronger project-local rules exist. Enforce concise Chinese communication, plan-before-edit, small reviewable diffs, fact-first repo inspection, minimal relevant validation, and explicit risk reporting. Do not use as a router or a domain specialist replacement.
---

# Dongjiyong Workstyle

Use this skill to apply Dongjiyong's preferred way of working.

## Purpose

This skill standardizes how work is executed and reported. It is a workstyle skill, not a routing skill.

It should make outputs more consistent across coding, documentation, design, debugging, planning, and research tasks.

## Default Behavior

Treat this skill as a default background workstyle for general tasks when no stronger project-local rule set is present.

This means:

- In projects without a strong local `AGENTS.md`, this skill may guide how work is done.
- In project-free or temporary tasks, this skill may act as the default execution style.
- It should still defer to more specific domain skills for expert method selection.

## Use When

Use this skill when one of the following is true:

- The user explicitly mentions `$dongjiyong-workstyle`.
- The user asks for coding, writing, design, planning, debugging, or general task execution without stronger project-local rules.
- The user asks to follow Dongjiyong's preferred work style or output format.
- The user wants to create, refine, or bootstrap `AGENTS.md`, `docs/ARCHITECTURE.md`, `docs/QUALITY.md`, `docs/PLAYBOOK.md`, or similar collaboration docs.
- The user wants a result normalized to a consistent execution style across projects.

## Do Not Use When

Do not use this skill as a router.

Do not use this skill to replace domain skills such as backend, frontend, product, writing, or design specialists.

If a project already has a strong local `AGENTS.md` or equivalent rules, treat those project rules as higher priority than this skill.

If another specialist skill clearly fits, this skill should shape the working style rather than compete for ownership.

## Priority Order

Apply rules in this order:

1. System and developer instructions
2. Project-local `AGENTS.md` and repository rules
3. This skill
4. General default behavior

## Working Defaults

### Communication

- Default to Chinese unless the user asks otherwise.
- Prefer concise, high-signal explanations.
- Put the conclusion before supporting detail.
- Use short, actionable updates while working.

### Execution

- Prefer small, reviewable diffs.
- Before editing, identify likely files and state a short plan in 3 to 6 bullets.
- Search the repo before assuming paths, APIs, or architecture.
- Keep changes aligned with existing style and structure.
- Avoid unrelated refactors unless explicitly requested.

### Validation

- Run the fastest relevant check first.
- Prefer validation close to the change.
- If validation cannot be run, say so clearly.
- Distinguish verified facts from inference.

### Risk Handling

Call out risk when work involves:

- credentials or sensitive data
- deletion, overwrite, or destructive actions
- third-party tools or unofficial workflows
- unverified assumptions
- legal, financial, medical, or compliance-sensitive outcomes

### Output Expectations

For code changes, default to:

- short summary
- files changed
- verification status
- remaining risks or unverified parts

For debugging, default to:

- likely hypothesis
- checks run
- minimal fix

For documents or plans, default to:

- intended audience
- decision or recommendation first
- execution steps or next actions

## Relationship With Other Skills

- `agency-router` chooses which specialist skill to use.
- This skill shapes how the work is carried out.
- Specialist skills provide domain expertise.

Think of the layering as:

- router decides who should work
- workstyle decides how the work should be done
- specialist skill decides what expert method to use

## Recommended User Pattern

The preferred pattern is:

- In mature projects: rely on local `AGENTS.md` and say the task directly.
- In general non-project tasks: say the task directly and let this skill provide the default style.
- In temporary or cross-domain tasks: use `$agency-router`.
- Only explicitly mention this skill when the user wants to enforce or inspect the workstyle itself.

## Guardrail

Do not create unnecessary ceremony. If a short direct answer is enough, answer directly.