Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计异常

kalopilotkalopilot 搜索

Agent Skill

kalopilot 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

2,619

周安装

107

GitHub Stars

2

下载量

839
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

2

许可证

unknown

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:kalopilot(kalopilot 搜索)
来源仓库:https://github.com/sailtonight/kalopilot-skill
仓库路径:skills/kalopilot
安装命令:
npx skills add https://github.com/sailtonight/kalopilot-skill --skill kalopilot
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/sailtonight/kalopilot-skill --skill kalopilot

简介

kalopilot 用于查找、检索和筛选相关信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词或任务场景快速定位候选结果。
  • 通过 npx skills add 命令从指定 GitHub 仓库安装并使用。
  • 安装前需确认权限范围、维护状态,以及是否会触发联网或文件操作。
  • 建议结合原始 README 核验具体用法和功能边界。

SKILL.md

KaloPilot — TikTok E-commerce Data Assistant

This skill lets you query TikTok Shop data through the KaloPilot AI agent. KaloPilot has access to 20+ specialized tools covering products, shops, creators, videos, livestreams, and categories across all TikTok Shop regions.

Authentication

Token is stored as plain text in ~/.kalopilot/token:

your-token-here

If the file doesn't exist, ask the user for their KaloData token, then save it:

mkdir -p ~/.kalopilot && echo -n "<token>" > ~/.kalopilot/token && chmod 600 ~/.kalopilot/token

API Endpoint

POST https://staging.kalodata.com/api/pilot/skill/ext/v1/chat/sync
Content-Type: application/json
Authorization: Bearer <token>

Making a Request

Use scripts/pilot.sh to manage queries. The script handles token loading, background execution, early error detection, and result cleanup automatically — preventing stale result issues.

Response time by complexity:

  • Simple queries (single-dimension lookup, e.g. "美国热门商品"): ~1 minute
  • Complex queries (cross-dimension, diagnostics, comparisons): 2–3 minutes
  • Very complex (multi-step analysis with reports): up to 10 minutes

Step 1 — Send query:

bash <skill-path>/scripts/pilot.sh query "<user question>"

With task_id for follow-up questions:

bash <skill-path>/scripts/pilot.sh query "<user question>" "<task_id>"

The script launches curl in the background, does a 2-second early error check (catches bad token, network issues), and prints the PID. Tell the user the query is running.

Step 2 — Poll for completion:

Poll based on query complexity:

  • Simple query → first poll after 45 seconds
  • Complex query → first poll after 90 seconds
  • If still running, poll again every 30 seconds
bash <skill-path>/scripts/pilot.sh status

Output is running, done, or No active query.

Step 3 — Read result:

Only call this after status shows done. The script refuses to read while a query is still running, and cleans up after reading — so stale results are never left behind.

bash <skill-path>/scripts/pilot.sh result

Multi-turn Conversations

The first request returns a task_id in the response. Always reuse that task_id for follow-up questions on the same topic — this gives KaloPilot conversation context so it can understand references like "the first one" or "compare it with yesterday".

Example flow:

  1. First message: {"query": "美国热门商品有哪些?"} → response includes "task_id": "abc123"
  2. Follow-up: {"query": "第一名的销售趋势怎么样?", "task_id": "abc123"}
  3. Another follow-up: {"query": "对比一下英国市场", "task_id": "abc123"}

When the user switches to a clearly different topic, start fresh without task_id.

Response Format

{
  "task_id": "abc123",
  "message_id": "456",
  "text": "The main analysis text...",
  "report": "# Detailed Report\n\n...",
  "report_url": "https://staging.kalodata.com/...",
  "token_usage": {...},
  "credits_consumed": 10
}

How to display the response

  1. Show the text field — this is the primary analysis.
  2. If report is not null, display it as well (it's a markdown report with tables and structured data).
  3. If report_url is not null, show it as a clickable link so the user can open the full report in their browser, e.g. "查看完整报告:[链接]". CRITICAL: Copy the report_url value verbatim from the API response. NEVER construct, guess, or fabricate this URL — the task_id and tool_call_id parameters in the URL are unique server-generated values that cannot be inferred. If report_url is absent or null in the response, do not show any link.
  4. Mention credits consumed at the end, e.g. "(consumed 10 credits)".

Error responses

The API returns errors as JSON with a message field. Common cases:

  • Insufficient credits: tell the user they need to top up on KaloData.
  • Timeout: suggest the user try a more specific query.
  • Other errors: show the error message directly.

What Users Can Ask

KaloPilot covers these dimensions across all TikTok Shop regions (US, UK, ID, MY, TH, VN, PH, SG, MX, DE, IT, FR, ES, BR, JP):

  • Products: rankings, sales trends, price analysis, breakout products, selling point analysis, comment/review analysis, image search (find similar products by photo)
  • Shops: store rankings, revenue, product mix, growth metrics
  • Creators/Influencers: performance, GMV, follower analysis, collaboration data, creator search
  • Videos: viral product videos, engagement metrics, conversion analysis, video script extraction
  • Livestreams: live sales data, viewer metrics, product performance during streams
  • Categories: market size, growth trends, competition landscape, category search by keyword
  • User's collections: view saved/followed products, shops, creators, videos, and video scripts
  • URL lookup: paste a TikTok, KaloData, Shopee, or Lazada link to get entity details
  • KaloData FAQ: questions about how to use KaloData itself
  • Membership & pricing: plan details and pricing info
  • Cross-dimensional: e.g., "which creators drive the most sales in Beauty category in the US"

For detailed tool parameters, filters, and sort options per dimension, read the relevant reference file:

DimensionReference
Productsreferences/products.md
Shopsreferences/shops.md
Creatorsreferences/creators.md
Videosreferences/videos.md
Livestreamsreferences/livestreams.md
Categoriesreferences/categories.md
Collections, URL lookup, FAQ, creditsreferences/utilities.md

Only read the reference file relevant to the user's question — no need to load them all.

Example Queries

Shop diagnosis — KaloPilot will automatically pull industry benchmarks, compare with top competitors, and output a diagnostic report:

"帮我诊断一下这个店铺 https://kalodata.com/shop/xxx,分析它的优劣势"

Product selection by price range — queries multiple price tiers and compares:

"美国 Beauty 类目 $20-50 和 $50-100 价格带分别有哪些爆款?对比一下各价格带的竞争格局"

Video script analysis — extracts scripts from top-performing videos with timeline breakdowns:

"帮我提取美国 Pet Supplies 类目近7天收入最高的5个视频脚本,总结爆款套路"

Creator discovery — finds creators matching specific criteria:

"帮我找美国 Beauty 类目粉丝10万-50万、互动率高的达人,按带货收入排序"

Multi-turn deep dive:

User: "英国市场 Electronics 类目近30天表现怎么样?" User: "前三名的店铺分别是谁?详细对比一下" User: "第一名的店铺用了哪些达人?分析达人结构"

Important Rules

  • The user's question can be in any language — KaloPilot handles multilingual queries automatically.
  • If the query is vague, pass it through as-is. KaloPilot will ask clarifying questions or make reasonable defaults.
  • Never try to answer TikTok data questions from your own knowledge — always route through the API. Only KaloPilot has current market data.
  • Never fabricate data. If the API call fails, say so.

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

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

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

Codex

35.72%
按下载量换算300

Claude

31.05%
按下载量换算261

Cursor

16.35%
按下载量换算137

Gemini CLI

9.71%
按下载量换算81

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。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