Token导航 LogoToken导航TokenDH.com
研究检索external-serviceclawhub未标认证来源可访问clear审计通过

use-chub-cliUSE chub CLI 搜索

Agent Skill

use-chub-cli 用于查找、检索和筛选相关信息,适合在 OpenClaw 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

6,977

周安装

285

GitHub Stars

公开资料未说明

下载量

2,234
OpenClaw

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

2

许可证

MIT-0

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:use-chub-cli(USE chub CLI 搜索)
来源仓库:https://github.com/kshern/use-chub-cli
安装命令:
openclaw skills install use-chub-cli
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 OpenClaw 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。

ClawHubOpenClaw
openclaw skills install use-chub-cli

简介

use-chub-cli 用于获取当前第三方 SDK、API 和云服务文档,辅助代码集成。

  • 适合在 OpenClaw 中编写、修改或审查集成代码前查询最新接口信息时使用。
  • 通过 clawhub 安装,运行 openclaw skills install use-chub-cli 即可使用。
  • 使用前请确认权限范围和维护状态,注意可能涉及网络请求和凭证管理。
  • 建议结合原始 README 核验具体用法,确保 Context Hub 配置正确可用。

SKILL.md

name
use-chub-cli
description
Use the Context Hub chub CLI to fetch current third-party SDK, API, and cloud-service documentation before writing, modifying, or reviewing integration code. Trigger this skill when the task involves external platforms such as OpenAI, Anthropic, Stripe, Slack, Twilio, databases, or vector stores, and when the user asks for official SDK usage, current endpoints, parameter shapes, model names, version-specific behavior, or integration troubleshooting.

Use chub CLI

Use chub to fetch current third-party documentation before writing code or answering API questions. Do not start chub-mcp unless the user explicitly asks for MCP.

Choose the command

  • Prefer the global command:
  chub --help
  • If chub is not on PATH, use a repository-local entrypoint when one exists. In this repository:
  node ./cli/bin/chub --help
  • Keep the command form consistent within the same task.

Workflow

1. Search for the right entry

Search by vendor first, then by capability. Use --json when the output needs to be consumed reliably.

chub search "openai"
chub search "stripe webhook" --json
chub search "pinecone vector db" --json

Follow these rules:

  • Prefer entries whose name, description, and tags all match the task.
  • If there are too many results, broaden once and then narrow.
  • If the user already provided an ID, run chub search <id> to confirm details.

2. Fetch the minimum necessary content

Fetch the entry point first. Pull additional files only when deeper detail is required.

chub get openai/chat --lang py
chub get stripe/api --lang js
chub get tavily/sdk --lang ts

Follow these rules:

  • Omit --lang only when the entry has a single language variant.
  • Use --file after the CLI shows Additional files available.
  • Use --full only when multiple reference files are genuinely needed.

3. Drill down selectively

chub get openai/chat --lang py --file references/streaming.md
chub get datahub/sdk --lang py --file references/search.md
chub get tavily/tavily-best-practices --full

Follow these rules:

  • Read the entry point before loading reference files.
  • Load only the files required to solve the current task.
  • Avoid --full for narrow questions.

4. Implement from the fetched docs

Follow these rules:

  • Treat chub get output as the source of truth instead of relying on memory.
  • Use the initialization pattern, method names, parameters, and version constraints shown in the docs.
  • If repository code conflicts with the docs, call out the conflict before choosing the safest implementation.
  • If chub has no matching entry, say so explicitly before falling back to official docs or user-provided material.

5. Capture local learning

If you discover a validated gotcha that the doc does not state clearly, save a local annotation:

chub annotate stripe/api "Webhook verification requires the raw request body before JSON parsing."

Capture only:

  • Project-specific constraints
  • Environment-specific differences
  • Version-specific gotchas
  • Validated conclusions missing from the docs

Do not capture:

  • Information already stated clearly in the doc
  • Unverified guesses
  • Long process notes

6. Ask before sending feedback

If the docs are clearly outdated or poor, suggest feedback, but do not send it without user approval:

chub feedback openai/chat up
chub feedback stripe/api down --label outdated

Decision rules

  • If the task asks for current SDK or API usage, run search and then get.
  • If the user gives only a vendor name, search first and then choose the closest capability match.
  • If the user wants a specific implementation language, include --lang.
  • If the task is about one detail, prefer --file over --full.
  • If the user dislikes MCP or the task only needs one-shot documentation lookup, stay with the CLI.

Failure handling

  • No results: retry once with a broader query.
  • Ambiguous matches: ask the user to confirm, or choose the best match based on the descriptions.
  • Language mismatch: switch to an available language or state that the needed language is missing.
  • chub not installed globally: use a repository-local entrypoint when available.

Quick examples

  • "Write a Python Responses API example with the OpenAI SDK"

Run chub search "openai" --json, then chub get openai/chat --lang py.

  • "Fix Stripe webhook verification errors"

Run chub search "stripe webhook" --json, then chub get stripe/api --lang js, and fetch reference files only if needed.

  • "Show how Tavily should be used in an agent workflow"

Run chub search "tavily" --json, then chub get tavily/tavily-best-practices.

Attribution

This skill is adapted from Context Hub and its bundled get-api-docs skill:

  • Repository: https://github.com/andrewyng/context-hub
  • Upstream skill: cli/skills/get-api-docs/SKILL.md
  • Original license: MIT

This version was modified for general-purpose publishing and reuse.

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

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

OpenClaw

86.36%
按下载量换算1,929

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

external-service

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills