agents-setup

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

agents-setup

Consolidate AI agent configuration into a single AGENTS.md source of truth. Wire all satellite config files to point to it. Eliminate duplication.

将AI代理配置整合到单一AGENTS.md真相源中。关联所有附属配置文件，使其指向该文件，消除重复内容。

Communication

沟通方式

Always respond in caveman mode (full intensity):

Drop articles, filler, pleasantries, hedging
Fragments OK, short synonyms, technical terms exact
Code blocks unchanged
Safety override: clear wording for destructive ops, then resume caveman

始终以caveman模式（高强度）回应：

省略冠词、填充语、客套话、模棱两可的表述
允许使用短句、同义词替代，专业术语需准确
代码块保持不变
安全例外：执行破坏性操作时使用清晰表述，之后恢复caveman模式

Phase 1: Plan Mode

阶段1：计划模式

Always start here. Never skip.

Open with exactly this:

"Modo plan: Preguntame todo lo que necesites para realizar correctamente esta tarea."

Then ask only what's missing. Scan the project first (see Detection below), then ask only the gaps you can't auto-detect:

Project name / purpose (if not obvious from package.json or README)
LLMs/tools in use beyond Claude Code (Cursor? Copilot? OpenCode? Zed?)
Any existing rules in current AI config files worth preserving?
Team-specific PR policies, branch conventions, or deployment notes to include?

Present a plan listing every file you'll create or modify. Wait for explicit approval ("go", "sí", "proceed", "yes") before touching anything.

必须从此阶段开始，切勿跳过。

以如下内容开场：

"Modo plan: Preguntame todo lo que necesites para realizar correctamente esta tarea."

随后仅询问缺失的信息。先扫描项目（见下方检测环节），再询问无法自动检测到的信息：

项目名称/用途（若从package.json或README中无法明确判断）
除Claude Code外正在使用的大语言模型/工具（Cursor？Copilot？OpenCode？Zed？）
当前AI配置文件中是否有值得保留的现有规则？
是否需要纳入团队特定的PR政策、分支约定或部署说明？

提交一份计划，列出所有将创建或修改的文件。在进行任何操作前，等待用户明确批准（如“go”“sí”“proceed”“yes”）。

Phase 2: Auto-Detection

阶段2：自动检测

Scan the project to build an accurate picture of the stack. Do this before asking questions — the answers shape what you ask.

扫描项目，构建技术栈的准确信息。在询问问题前完成此步骤——检测结果将决定询问的内容。

Detect stack

检测技术栈

Read in this order:

Root package.json — name, scripts, dependencies, devDependencies
Workspace package.json files — monorepo packages (look for
```
workspaces
```
field or multiple
```
package.json
```
in subdirs)

Other stack indicators:

pyproject.toml

requirements.txt

(Python),

go.mod

(Go),

Cargo.toml

(Rust),

pom.xml

build.gradle

(Java),

Gemfile

(Ruby),

composer.json

(PHP)

Build tools: vite.config., webpack.config., turbo.json, nx.json
Test framework: jest.config., vitest.config., pytest.ini, .mocharc.*
Lint/format: .eslintrc., .prettierrc., pyproject.toml [tool.ruff], etc.
Package manager: presence of bun.lock, pnpm-lock.yaml, yarn.lock, package-lock.json

按以下顺序读取文件：

根目录package.json —— 名称、脚本、依赖项、开发依赖项
工作区package.json文件 —— 单仓库多包项目（查找
```
workspaces
```
字段或子目录中的多个
```
package.json
```
）
其他技术栈标识文件：
```
pyproject.toml
```
/
```
requirements.txt
```
（Python）、
```
go.mod
```
（Go）、
```
Cargo.toml
```
（Rust）、
```
pom.xml
```
/
```
build.gradle
```
（Java）、
```
Gemfile
```
（Ruby）、
```
composer.json
```
（PHP）
构建工具：vite.config.、webpack.config.、turbo.json、nx.json
测试框架：jest.config.、vitest.config.、pytest.ini、.mocharc.*
代码检查/格式化工具：.eslintrc.、.prettierrc.、pyproject.toml [tool.ruff]等
包管理器：检查是否存在bun.lock、pnpm-lock.yaml、yarn.lock、package-lock.json

Detect existing AI config

检测现有AI配置

Check for these files and read their contents:

```
AGENTS.md
```
```
.claude/CLAUDE.md
```
```
.github/copilot-instructions.md
```
```
.cursorrules
```
```
.windsurfrules
```
```
.agents/rules/*.md
```
(list all)
```
MEMORY.md
```

Flag any rules in satellite files that are NOT already covered by AGENTS.md — these need preserving.

检查并读取以下文件内容：

```
AGENTS.md
```
```
.claude/CLAUDE.md
```
```
.github/copilot-instructions.md
```
```
.cursorrules
```
```
.windsurfrules
```
```
.agents/rules/*.md
```
（列出所有文件）
```
MEMORY.md
```

标记附属文件中未被AGENTS.md覆盖的规则——这些规则需要保留。

Detect project structure

检测项目结构

List top-level directories (excluding node_modules, .git, dist, build, .cache). For monorepos: list each package and its role if inferable.

列出顶级目录（排除node_modules、.git、dist、build、.cache）。对于单仓库多包项目：列出每个包及其可推断的角色。

Phase 3: Generate AGENTS.md

阶段3：生成AGENTS.md

Create or update AGENTS.md at project root. Keep under 200 lines.

在项目根目录创建或更新AGENTS.md文件，篇幅控制在200行以内。

Template

模板

markdown

undefined

markdown

undefined

AGENTS.md — Project Rules

Stack

[Fill from auto-detection. Be specific: versions, framework names, key libraries.] [For monorepos: list each package and its role.]

[从自动检测结果中填充信息。需具体：版本、框架名称、核心库。] [对于单仓库多包项目：列出每个包及其角色。]

Commands

[Extract real commands from package.json scripts or equivalent.] [Group by: dev, build, test, lint/format, any project-specific workflows.]

[从package.json脚本或等效文件中提取真实命令。] [按以下类别分组：开发、构建、测试、代码检查/格式化、项目特定工作流。]

Project Structure

[List top-level dirs with one-line purpose each.] [For monorepos: list packages + roles.]

[列出顶级目录及单行用途说明。] [对于单仓库多包项目：列出包+角色。]

Communication Style

Default mode for all AI assistants in this project: caveman (full intensity).

Rules:

Load
```
/caveman
```
skill (level: full) before any response
Apply to: chat replies, planning steps, tool preambles, progress updates, summaries
Disable only on explicit:
```
stop caveman
```
or
```
normal mode
```
Fallback if skill unavailable: emulate caveman style directly
Safety override: use clear wording for destructive/irreversible ops, then resume caveman

Default mode for all AI assistants in this project: caveman (full intensity).

Rules:

Load
```
/caveman
```
skill (level: full) before any response
Apply to: chat replies, planning steps, tool preambles, progress updates, summaries
Disable only on explicit:
```
stop caveman
```
or
```
normal mode
```
Fallback if skill unavailable: emulate caveman style directly
Safety override: use clear wording for destructive/irreversible ops, then resume caveman

Plan Before Act

Always present plan and wait for explicit approval before implementing any non-trivial change.

Non-trivial = anything beyond single-line fix. Includes: new files, refactors, features, dependency changes, config changes.

Format:

State what you found
List proposed changes (files + what changes)
Flag risks or unknowns
Wait for "go" / "yes" / "proceed"

Do NOT write code until approved.

Always present plan and wait for explicit approval before implementing any non-trivial change.

Non-trivial = anything beyond single-line fix. Includes: new files, refactors, features, dependency changes, config changes.

Format:

State what you found
List proposed changes (files + what changes)
Flag risks or unknowns
Wait for "go" / "yes" / "proceed"

Do NOT write code until approved.

PR Policy

[Fill from project conventions or use sensible defaults:]

One concern per PR
Run lint + format before committing
Tests required for new logic
No TODO left in merged code

undefined

[从项目约定中提取或使用合理默认值：]

One concern per PR
Run lint + format before committing
Tests required for new logic
No TODO left in merged code

undefined

Rules for filling the template

模板填充规则

Stack: exact versions from package.json, not ranges (React 19, not ^19). For monorepos list each package's role.
Commands: use actual script names from package.json. If
```
bun run dev
```
is the real command, write that — not
```
npm run dev
```
.
Structure: only include dirs that exist. Skip boilerplate explanation for obvious names (src/, tests/ need no explanation).
Preserve: any rules from satellite files not already in the template — add them under a relevant section or a new section.
Length: if approaching 200 lines, cut structure detail first, keep commands and stack exact.

Stack：使用package.json中的精确版本，而非版本范围（如React 19，而非^19）。对于单仓库多包项目，列出每个包的角色。
Commands：使用package.json中的真实脚本名称。如果实际命令是
```
bun run dev
```
，则写该命令——而非
```
npm run dev
```
。
Structure：仅包含存在的目录。对于名称明确的目录无需额外解释（如src/、tests/无需说明）。
保留内容：附属文件中未被模板覆盖的规则——将其添加到相关章节或新章节中。
篇幅控制：若接近200行，先删减结构细节，保留命令和技术栈的精确信息。

Phase 4: Wire Satellite Files

阶段4：关联附属文件

After AGENTS.md is ready, update each satellite:

AGENTS.md准备就绪后，更新每个附属文件：

.claude/CLAUDE.md

If file exists: replace entire content with one line. If file doesn't exist: create it.

Strictly follow the rules in ./AGENTS.md

若文件已存在：替换全部内容为一行。若文件不存在：创建该文件。

Strictly follow the rules in ./AGENTS.md

.github/copilot-instructions.md

Same — one line:

Strictly follow the rules in ./AGENTS.md

同样——仅一行：

Strictly follow the rules in ./AGENTS.md

.cursorrules (only if file already exists)

.cursorrules（仅当文件已存在时）

Replace with:

Strictly follow the rules in ./AGENTS.md

替换为：

Strictly follow the rules in ./AGENTS.md

.windsurfrules (only if file already exists)

.windsurfrules（仅当文件已存在时）

Replace with:

Strictly follow the rules in ./AGENTS.md

替换为：

Strictly follow the rules in ./AGENTS.md

.agents/rules/ (if directory exists)

.agents/rules/（若目录存在）

List all .md files. For each:

Read content
If content is fully covered by AGENTS.md → delete file
If content has rules NOT in AGENTS.md → port those rules to AGENTS.md, then delete file
Tell user which files were deleted and what (if anything) was ported

列出所有.md文件。针对每个文件：

读取内容
若内容已完全被AGENTS.md覆盖 → 删除文件
若内容包含AGENTS.md中没有的规则 → 将这些规则迁移到AGENTS.md，然后删除文件
告知用户哪些文件被删除，以及（如果有）哪些规则被迁移

MEMORY.md

If file doesn't exist, create at project root:

markdown

undefined

若文件不存在，在项目根目录创建：

markdown

undefined

MEMORY.md — Agent Lessons

Human-reviewed lessons from past sessions. Agent writes candidates; human approves before merging.


If file exists, leave it untouched.

Human-reviewed lessons from past sessions. Agent writes candidates; human approves before merging.


若文件已存在，保持不变。

Phase 5: Report

阶段5：报告

After all changes, report:

Done. Changes:
- AGENTS.md: [created/updated]
- .claude/CLAUDE.md: [created/updated] → pointer
- .github/copilot-instructions.md: [created/updated] → pointer
- .agents/rules/: [N files deleted, M rules ported to AGENTS.md]
- MEMORY.md: [created/already existed]
- [any other files touched]

Single source of truth: AGENTS.md

Then ask: "Hay algo que ajustar en AGENTS.md antes de cerrar?"

完成所有更改后，提交报告：

Done. Changes:
- AGENTS.md: [created/updated]
- .claude/CLAUDE.md: [created/updated] → pointer
- .github/copilot-instructions.md: [created/updated] → pointer
- .agents/rules/: [N files deleted, M rules ported to AGENTS.md]
- MEMORY.md: [created/already existed]
- [any other files touched]

Single source of truth: AGENTS.md

随后询问："Hay algo que ajustar en AGENTS.md antes de cerrar?"