Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问clear审计异常

archiarchi 搜索

Agent Skill

archi 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

196

周安装

8

GitHub Stars

公开资料未说明

下载量

63
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:archi(archi 搜索)
来源仓库:https://github.com/endlesshoper/archi
仓库路径:skills/archi
安装命令:
npx skills add https://github.com/endlesshoper/archi --skill archi
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/endlesshoper/archi --skill archi

简介

archi 提供 AI 辅助开发的脚手架与项目管理流程,支持任务分解、特性管理与文档同步。

  • 适合启动新项目或跟踪进行中功能,通过命令 /archi 自动识别上下文并推荐下一步行动。
  • 使用时可初始化项目结构、添加特性卡片或查看状态概览,所有操作均围绕清晰的工作流展开。
  • 安装前请确认终端环境支持交互式命令输入,注意该技能依赖本地状态文件持久化,需保留 ~/.openclaw 目录权限。
  • archi 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Archi

Project scaffolding and management for AI-assisted development ("vibe coding").

Workflows

CommandWhat it does
/archiDetect context and choose appropriate workflow, or show menu
/archi initInitialize new project with docs structure
/archi featureManage features (add/start/complete/update)
/archi statusShow project overview
/archi docsSync and update documentation

Users can just run /archi - you'll figure out what they need based on context. If it's a new project with no docs structure, offer to initialize. If they're mid-conversation about a feature, offer to track it. When unclear, use AskUserQuestion to let them choose.

Parse $ARGUMENTS to route to specific workflows, or detect from context.


Initialize Project

Trigger: /archi init, /archi start, or when project has no docs structure

Goal: Create SPEC.md and docs folder structure through a conversational interview.

Interview the user to understand:

  • Project name (detect from directory/package.json if possible)
  • What they're building (one sentence)
  • Who it's for
  • What problem it solves
  • Key features (3-5 things it should do)

If there are existing files (README, package.json, etc.), read them to gather context before asking questions. Don't ask permission to read - just read them.

Use AskUserQuestion throughout to gather info interactively. Keep questions simple and beginner-friendly.

Create this structure:

project/
├── SPEC.md
└── docs/
    ├── features/
    │   ├── todo/
    │   ├── doing/
    │   └── done/
    ├── research/
    │   ├── spikes/
    │   └── references/
    └── bugs/

SPEC.md template:

# {Project Name}

> {One-sentence description}

## Problem
{What problem does this solve}

## Users
{Who is this for}

## Features
1. {Feature}
2. {Feature}
...

## Out of Scope
- TBD

## Success Criteria
- TBD

After creating, suggest next steps like adding their first feature.


Feature Management

Trigger: /archi feature, or when user mentions working on/completing something

Goal: Track features through todo → doing → done lifecycle.

Capabilities:

  • Add new feature → create file in docs/features/todo/
  • Start feature → move from todo/ to doing/, add started date
  • Complete feature → move from doing/ to done/, add completed date
  • Update progress → edit existing feature file

Detect intent from conversation context. If user says "I just finished the auth system", offer to mark it done. If they say "I'm going to work on payments", offer to start that feature or create it if it doesn't exist.

Feature file format:

Todo:

---
title: Feature Name
priority: high | medium | low
---
# Feature Name

Description and requirements...

Doing:

---
title: Feature Name
priority: high
started: 2026-01-23
---
# Feature Name

## Progress
- What's been done
- What's next

Done:

---
title: Feature Name
category: Core | Enhancement | Fix | Infrastructure
completed: 2026-01-23
---
# Feature Name

Summary of what was built.

Use kebab-case for filenames: user-authentication.md


Project Status

Trigger: /archi status, or when user asks about project state

Goal: Give a quick overview of where the project stands.

Read the docs structure and summarize:

  • Features in progress (from doing/)
  • Backlog count and priorities (from todo/)
  • Recently completed (from done/)
  • Any research or bugs

Keep it concise and actionable.


Documentation Sync

Trigger: /archi docs, or when docs seem out of sync

Goal: Keep project documentation current.

Check for:

  • SPEC.md reflects current scope
  • Features mentioned but not tracked
  • Stale information

Offer to update what's outdated. Don't over-ask - if something clearly needs updating, just suggest the specific changes.


Behavior Guidelines

  1. Be contextual - detect what the user needs from conversation flow
  2. Be beginner-friendly - no jargon, explain what's happening
  3. Just do it - read files, check structure, don't ask permission for basic operations
  4. Use AskUserQuestion - for choices and gathering info, not for confirmation of obvious actions
  5. Kebab-case filenames - user-authentication.md not UserAuthentication.md
  6. Date format - YYYY-MM-DD

Handling Existing Documentation

IMPORTANT: Never overwrite existing documentation without explicit user permission.

When initializing a project that already has documentation (README.md, docs/, SPEC.md, etc.):

  1. Read existing docs first - understand what's already there
  2. Preserve all content - never summarize or condense existing documentation
  3. Adapt to archi format - restructure into the archi structure while keeping ALL original content
  4. Ask before overwriting - if a file would be replaced, ask the user first
  5. Copy when needed - if existing content fits elsewhere, use cp to preserve it before reorganizing

Migration approach:

  • Existing README → Extract relevant sections into SPEC.md, keep README for public-facing info
  • Existing docs/ folder → Move files into appropriate archi subdirectories (features/, research/, etc.)
  • Existing feature docs → Adapt to feature file format, preserving full content
  • Existing specs → Merge into SPEC.md sections, keeping all detail

Never lose user content. If unsure where something belongs, ask. If content is substantial, copy the file rather than moving it until the user confirms the new structure works for them.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

25.43%
按下载量换算16

windsurf

23.61%
按下载量换算15

OpenCode

16.83%
按下载量换算11

Codex

12.23%
按下载量换算8

Antigravity

7.39%
按下载量换算5

Gemini CLI

3.16%
按下载量换算2

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills