Token导航 LogoToken导航

Agent Skill · 开发

builderbot-code-skill(builderbot 代码技能)

builderbot-code-skill 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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。

builderbot-code-skill 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描有提醒项,按安装量在开发类的 21,746 个 Skill 里排第 11,062。

总安装 1,956周安装 84GitHub Stars 4下载 685核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:builderbot-code-skill(builderbot 代码技能)
来源仓库:https://github.com/leifermendez/skill-builderbot-code
仓库路径:skills/builderbot-code-skill
安装命令:
npx skills add https://github.com/leifermendez/skill-builderbot-code --skill builderbot-code-skill
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。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/leifermendez/skill-builderbot-code --skill builderbot-code-skill

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。适合围绕仓库状态、代码变更或协作事项进行整理。可结合来源仓库和原始 README 核验具体用法。安装前建议确认权限范围、维护状态及是否触发联网或文件读写。

builderbot-code-skill 属于开发类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/builderbot-code-skill

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/leifermendez/skill-builderbot-code --skill bui…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

builderbot-code-skill 怎么安装?

在终端执行 npx skills add https://github.com/leifermendez/skill-builderbot-code --skill builderbot-co…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

builderbot-code-skill 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

builderbot-code-skill 需要什么权限?

权限级别是只读。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。

builderbot-code-skill 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

builderbot-code-skill 用的人多吗?

公开数据是累计安装 1,956 次,近一周安装 84 次,下载 685 次,GitHub 4 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

builderbot-code-skill 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
builderbot-code-skill

builderbot 代码技能 · 本页

1,9564只读未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

BuilderBot Code Skill

Operate

  1. Identify the stack — confirm provider package, database adapter, and flow entrypoints before writing code.
  2. Type the callback signature correctly:

- ctx: BotContext → {body, from, name, host,...platform fields} - methods: BotMethods → {flowDynamic, gotoFlow, endFlow, fallBack, state, globalState, blacklist, provider, database, extensions}

  1. Enforce flow-control semantics — these three MUST be returned:

- return gotoFlow(flow) · return endFlow('msg') · return fallBack('msg') - These MUST be awaited: await flowDynamic(...) · await state.update(...) · await globalState.update(...)

  1. Prefer small, composable flows — one responsibility per flow file.
  2. Keep secrets in env vars — never hardcode tokens, keys, or credentials.
  3. After writing any flow, run through the UX Review Checklist below — simulate the conversation as the end-user and verify every item before considering the flow done.
  4. Always validate lint before finishing — run eslint. --no-ignore and fix every error before delivering code.

Critical Rules (common bugs)

BugWrongRight
Missing return on flow controlgotoFlow(flow)return gotoFlow(flow)
Missing await on state/dynamicstate.update({...})await state.update({...})
flowDynamic with raw string arrayflowDynamic(['a','b','c']) — sends 3 separate messagesflowDynamic([{body: ['a','b','c'].join('\n'), delay: rnd()}])
flowDynamic + endFlow in same callbackawait flowDynamic(...); return endFlow()Split into two chained addAction — first sends flowDynamic, second calls return endFlow(...)
Unnecessary dynamic importawait import('./flow') when no circular dependency existsTop-level import {flow} from './flow' — only use dynamic import when flow A → B AND B → A
Circular ESM importTop-level import between flows that reference each otherDynamic const {xFlow} = await import('./x.flow') inside the callback
idle without capture{idle: 5000}{capture: true, idle: 5000}
Using require() in ESMrequire('./flow')await import('./flow') — project is "type": "module", require is not available
Using buttons{buttons: [...]}Text-based numbered menu + capture: true
Checking idleFallBackif (ctx?.idleFallBack) {...}Never use idleFallBack — just set {capture: true, idle: N} and let the flow expire automatically

Flow Chain

addKeyword(keywords, options?)        // ActionPropertiesKeyword: capture, idle, media, delay, regex, sensitive  ← NEVER use `buttons`
  .addAnswer(message, options?, cb?, childFlows?)
  .addAction(cb)
// cb: async (ctx: BotContext, methods: BotMethods) => void

State Quick Reference

// Per-user
await state.update({ key: value })
state.get('key')            // dot notation supported: 'user.profile.name'
state.getMyState()
state.clear()

// Global (shared across all users)
await globalState.update({ key: value })
globalState.get('key')
globalState.getAllState()

UX Review Checklist

After building any flow, mentally walk through the conversation as the end-user and verify every point below. Fix anything that fails before delivering the code.

#QuestionWhat to check / fix
1Does every prompt tell the user exactly what to type?Each addAnswer with capture: true must state the expected input (e.g. "Reply with *1*, *2*, or *3*").
2Are all invalid inputs handled?Every captured step must have a fallBack('...') branch for bad input.
3Is there an idle timeout on long captures?Add {capture: true, idle: 60000} — the flow will automatically expire after the timeout.
4Can the user always exit?Provide a cancel keyword (e.g. "cancel", "salir") or honour it inside captures and call return endFlow('...').
5Are messages short, mobile-friendly, and max 3-4 bubbles?No wall-of-text AND no bubble spam. Group lines with \n inside one FlowDynamicMessage.body. Each string in a flowDynamic([...]) array = a separate WhatsApp message — never spread long lists. Add random delay per bubble.
6Is the user's name used where natural?Greetings and confirmations should reference ctx.name when available.
7Are menus numbered text lists (never buttons)?Use ['Option 1', '1. Foo', '2. Bar'] + capture: true + fallBack.
8Does each multi-step flow confirm before committing?Before irreversible actions (order, payment, delete) show a summary and ask "confirm? *yes / no*".
9Does the flow end with a clear closing message?The final step must tell the user what happened and what to do next (or say goodbye).
10Are error messages actionable?Never say just "error". Say what went wrong and how to fix it: return fallBack('That doesn\'t look like a valid email. Try again:')

WhatsApp Text Formatting

WhatsApp uses its own markdown — always apply it in message strings.

StyleSyntaxExample
Bold*text**Pepperoni*
Italic_text__Tomate y mozzarella_
Strikethrough~text~~$15~
Monospace``` `text` `````` `CODE123` ```
Bullet list• or -• Opción 1

Rules

  • Use *bold* for product names, totals, section headers, and actions the user must take.
  • Use _italic_ for descriptions, hints, and secondary info.
  • Use • (not -) for list items — renders cleaner on mobile.
  • Use ━━━━━━━ (or ---) as a visual divider between sections inside one bubble.
  • Never use HTML tags (<b>, <br>, etc.) — WhatsApp ignores them.
  • Never use standard markdown (**bold**, ## heading) — not supported.

Example — well-formatted bubble

const body = [
    '*🍕 Tu pedido*',
    '━━━━━━━━━━━━━━',
    `• Pizza: *Pepperoni*`,
    `• Tamaño: *Mediana (30 cm)*`,
    `• Cantidad: *2*`,
    '',
    `💰 Total: *$26 USD*`,
    '',
    '_Responde *sí* para confirmar o *no* para cancelar._',
].join('\n')

WhatsApp Messaging Norms

These rules apply to every flow. Violating them makes the bot feel like spam.

Critical distinction — flowDynamic vs addAnswer with arrays:

CallArray behavior
flowDynamic(['a','b','c'])⚠️ Each string = separate WhatsApp message
addAnswer(['a','b','c'])✅ Joined into one message with line breaks
Always use FlowDynamicMessage[] with .join('\n') in body when calling flowDynamic. Never pass raw string arrays.
RuleWrongRight
Max 3-4 bubbles per turnflowDynamic(['line1','line2','line3',...]) → each string = 1 bubbleflowDynamic([{body: ['line1','line2','line3'].join('\n'), delay: rnd()}])
Always use random delayNo delay between bubblesdelay: Math.floor(Math.random() * 800) + 500 on each bubble
Never spread item listsflowDynamic([...items.map(...)])flowDynamic([{body: items.map(...).join('\n'), delay: rnd()}])

Random delay helper (define once per file)

const rnd = () => Math.floor(Math.random() * 800) + 500

Correct pattern — 3 bubbles max

await flowDynamic([
    {
        body: [
            '*Título*',
            '',
            items.map(i => `• ${i.name} — $${i.price}`).join('\n'),
        ].join('\n'),
        delay: rnd(),
    },
    {
        body: '*Sección 2*\nLínea A\nLínea B',
        delay: rnd(),
    },
])
// addAnswer o addAction siguiente = bubble 3

Presence Update (Baileys / Sherpa only)

Simulates "typing..." or "recording..." before sending a message. Makes the bot feel human. Only available with BaileysProvider and SherpaProvider.
type WAPresence = 'unavailable' | 'available' | 'composing' | 'recording' | 'paused'
// composing = typing bubble   recording = audio bubble

Pattern — typing indicator before each message

const waitT = (ms: number) => new Promise(resolve => setTimeout(resolve, ms))

.addAction(async (ctx, { provider, flowDynamic }) => {
    await provider.vendor.sendPresenceUpdate('composing', ctx.key.remoteJid)
    await waitT(1500)
    await flowDynamic([{ body: 'Mensaje que parece escrito por humano', delay: rnd() }])
    await provider.vendor.sendPresenceUpdate('paused', ctx.key.remoteJid)
})

Rules

  • Always call sendPresenceUpdate('paused',...) after sending — clears the indicator.
  • Combine with rnd() delays: presence update → wait → flowDynamic → paused.
  • Use composing for text replies, recording for audio context.
  • Never use on non-Baileys/Sherpa providers — will throw at runtime.

Debug Checklist

  • Flow not switching → return gotoFlow(...)
  • Session not ending → return endFlow(...)
  • Fallback not repeating → return fallBack(...)
  • State not saved → await state.update(...)
  • Idle not firing → add capture: true alongside idle (never check idleFallBack)
  • EVENTS flow not triggering → verify provider maps the event payload to EVENTS.*
  • Circular import crash → use dynamic await import() inside the callback (never require())
  • Language server shows "Cannot find module" on dynamic imports → check if a circular dep actually exists; if not, replace with static top-level import
  • ESLint errors after changes → run eslint. --no-ignore and fix before finishing
  • builderbot/func-prefix-endflow-flowdynamic error → flowDynamic and endFlow are in the same callback; split into two chained addAction: .addAction(async (_, {flowDynamic}) => {await flowDynamic(lines)}).addAction(async (_, {endFlow}) => {return endFlow('bye')})

Module System

This project uses ESM ("type": "module" in package.json, "module": "ES2022" in tsconfig).

  • NEVER use require() — it is not available in ESM.
  • Default: use static top-level imports. Dynamic imports cause TypeScript language server errors when no circular dependency exists.
  • Before using await import(), draw the dependency graph. Dynamic import is only justified when flow A calls flow B and flow B calls flow A (a real cycle).
✅ Static (no cycle):  welcome → menu → order → payment
   import { orderFlow } from './order.flow'   ← top of file

✅ Dynamic (real cycle): welcome ↔ order
   const { orderFlow } = await import('./order.flow')  ← inside callback

When flows reference each other (circular), use dynamic import() inside the callback:

.addAction(async (ctx, { gotoFlow }) => {
    const { targetFlow } = await import('./target.flow')
    return gotoFlow(targetFlow)
})

Modular Structure (recommended)

Organize flows in a src/flows/ directory with a barrel index.ts that exports the assembled flow:

src/
├── app.ts
├── flows/
│   ├── index.ts            # createFlow([...all flows])
│   ├── welcome.flow.ts
│   └── order.flow.ts
└── services/
// src/flows/index.ts
import { createFlow } from '@builderbot/bot'
import { welcomeFlow } from './welcome.flow'
import { orderFlow } from './order.flow'

export const flow = createFlow([welcomeFlow, orderFlow])
// src/app.ts
import { createBot, createProvider } from '@builderbot/bot'
import { MemoryDB as Database } from '@builderbot/bot'
import { BaileysProvider as Provider } from '@builderbot/provider-baileys'
import { flow } from './flows'

const main = async () => {
    const provider = createProvider(Provider)
    const database = new Database()
    await createBot({ flow, provider, database })
    provider.initHttpServer(+(process.env.PORT ?? 3008))
}
main()

References

  • Code patterns (scaffold, capture, gotoFlow, media, idle, flowDynamic, fallBack, REST, EVENTS, UX patterns): patterns.md
  • Provider configs, database configs, TypeScript types: providers.md

平台分布

Codex

34.34%
按下载量换算235

Claude

29.3%
按下载量换算201

Cursor

19.64%
按下载量换算135

Gemini CLI

8.34%
按下载量换算57

安全审计

Gen Agent Trust Hub

可疑

Socket

通过

Snyk

可疑

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills