Back to Details

tiktok-shop-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

TikTok Shop Research

TikTok Shop 研究

Follow shared release-shell rules in:
  • postplus-shared
    release-shell rules
Use this skill when the user wants TikTok Shop data gathered, normalized, and analyzed locally.
This skill is not a fixed workflow for one narrow task. Its main job is to help the agent:
  • interpret flexible TikTok Shop research requests
  • choose a suitable hosted collection key
  • map the request into hosted collection input fields
  • normalize raw output into one stable local shape
  • run lightweight analysis without binding the workflow to one product or one report format
遵循以下共享release-shell规则：
  • postplus-shared
    release-shell规则
当用户需要本地采集、标准化并分析TikTok Shop数据时，使用此技能。
本技能并非针对单一窄任务的固定工作流。其主要作用是帮助Agent：
  • 解读灵活的TikTok Shop研究请求
  • 选择合适的hosted collection key
  • 将请求映射为hosted collection输入字段
  • 将原始输出标准化为统一的本地格式
  • 运行轻量分析，且不将工作流绑定到单一产品或报告格式

Core Principle

核心原则

The user will usually describe a research goal, not an actor input schema.
Your job is to translate the user request into:
  • scrape target type
  • hosted collection key
  • input JSON
  • local output paths
  • optional normalization
  • optional analysis
Do not expose raw actor complexity unless it matters.
用户通常会描述研究目标，而非Actor输入模式。
你的任务是将用户请求转换为：
  • 抓取目标类型
  • hosted collection key
  • 输入JSON
  • 本地输出路径
  • 可选的标准化操作
  • 可选的分析操作
除非必要，否则不要暴露原始Actor的复杂性。

Supported Task Shapes

支持的任务类型

This skill is suitable for requests such as:
  • analyze one product page
  • compare several product URLs
  • inspect one shop
  • sample a category or result set
  • summarize pricing, shops, reviews, or benchmark candidates
  • collect a local dataset first and analyze later
The user does not need to know which actor is being used.
本技能适用于以下类型的请求：
  • 分析单个商品页面
  • 对比多个商品URL
  • 调研单个店铺
  • 抽样某个品类或结果集
  • 总结定价、店铺、评论或标杆候选商品信息
  • 先采集本地数据集，再进行后续分析
用户无需了解使用的是哪个Actor。

Decision Model

决策模型

Before running anything, classify the request into one of these shapes:
在执行任何操作前，将请求归类为以下类型之一：

1. Product Lookup

1. 商品查询

Use when the user gives:
  • one product URL
  • several product URLs
  • "analyze this listing"
Preferred hosted collection key:
  • tiktok-shop-products
当用户提供以下内容时使用：
  • 单个商品URL
  • 多个商品URL
  • “分析这个Listing”
首选hosted collection key：
  • tiktok-shop-products

2. Shop Audit

2. 店铺审计

Use when the user asks:
  • "look at this shop"
  • "what is this seller pushing"
  • "compare these shops"
Preferred hosted collection key:
  • tiktok-shop-products
当用户提出以下请求时使用：
  • “看看这家店铺”
  • “这个卖家主推什么商品”
  • “对比这些店铺”
首选hosted collection key：
  • tiktok-shop-products

3. Category or Search Sampling

3. 品类或搜索抽样

Use when the user asks:
  • "what products show up for this category"
  • "what price bands show up in this niche"
  • "find benchmark products in this segment"
Use 
tiktok-shop-products
when the request can be expressed as TikTok Shop product, shop, or category URLs.
If the request cannot be expressed as URLs for the released collection key, stop and state that constraint.
当用户提出以下请求时使用：
  • “这个品类下有哪些商品”
  • “这个细分领域的价格区间是怎样的”
  • “在这个细分市场中找到标杆产品”
如果请求可以转化为TikTok Shop商品、店铺或品类URL，使用
tiktok-shop-products
。
如果请求无法转化为已发布的collection key对应的URL，请停止并说明该限制。

Hosted Collection Strategy

Hosted Collection策略

Current defaults:
  • tiktok-shop-products
Use the hosted collection key by default because it is the released product-shell path.
Do not switch to unreleased collection paths when:
  • the hosted collection fails
  • the page type is unsupported
  • the user asks for a query shape that the released collection key does not support
Treat provider selection as server-owned infrastructure.
当前默认设置：
  • tiktok-shop-products
默认使用hosted collection key，因为它是已发布的product-shell路径。
在以下情况中，请勿切换到未发布的collection路径：
  • hosted collection执行失败
  • 页面类型不被支持
  • 用户请求的查询类型是已发布的collection key不支持的
将提供商选择视为服务器所属的基础设施。

Input Mapping Rule

输入映射规则

Map the user request into hosted collection input in the lightest possible way.
Typical mappings:
  • single product URL -> 
    start_urls: [{ "url": "https://..." }]
  • multiple product, shop, or category URLs -> 
    start_urls
  • optional sample bound -> 
    max_items
Build the smallest valid input for 
tiktok-shop-products
.
When the user request is vague, prefer:
  1. a small sample
  2. URL-driven collection if URLs are available
  3. normalization after scraping
  4. analysis after normalization
以最简方式将用户请求映射为hosted collection输入。
典型映射：
  • 单个商品URL -> 
    start_urls: [{ "url": "https://..." }]
  • 多个商品、店铺或品类URL -> 
    start_urls
  • 可选抽样限制 -> 
    max_items
为
tiktok-shop-products
构建最小的有效输入。
当用户请求模糊时，优先选择：
  1. 小样本量
  2. 若有URL则基于URL采集
  3. 抓取后进行标准化
  4. 标准化后进行分析

Output Contract

输出约定

When raw scraping is performed, prefer this progression:
  1. raw actor output
  2. normalized dataset
  3. optional analysis summary
Normalized datasets follow:
  • schemas/tiktok-shop-dataset.schema.json
The normalized shape exists so downstream analysis does not depend on actor-specific fields.
Each normalized item should preserve:
  • source metadata
  • product identity
  • shop identity
  • pricing
  • demand / proof signals
  • merchandising details
  • media
  • raw original payload
执行原始抓取时，优先遵循以下流程：
  1. 原始Actor输出
  2. 标准化数据集
  3. 可选的分析摘要
标准化数据集遵循：
  • schemas/tiktok-shop-dataset.schema.json
标准化格式的存在是为了让下游分析不依赖于Actor特定的字段。
每个标准化条目应保留：
  • 源元数据
  • 商品标识
  • 店铺标识
  • 定价信息
  • 需求/验证信号
  • 商品详情
  • 媒体内容
  • 原始载荷

Release-Shell Execution Contract

Release-Shell执行约定

  • keep actor input JSON, raw datasets, normalized datasets, and analysis caches under 
    <work-folder>/.postplus/tiktok-shop/
  • keep only final user-facing summaries or shortlisted exports outside 
    .postplus/
  • compile the user's intent into a small actor-input file before the expensive collection step
  • start with a bounded first pass:
    • one URL
    • one shop
    • one small category sample
  • if hosted capability is unavailable, unauthorized, or returns a stable network error, stop immediately instead of switching to ad hoc shell glue
If the user does not specify paths, choose clear descriptive names under 
.postplus/
for intermediates and make the final output path explicit.
  • 将Actor输入JSON、原始数据集、标准化数据集及分析缓存存储在
    <work-folder>/.postplus/tiktok-shop/
    目录下
  • 仅将最终面向用户的摘要或精选导出内容存储在
    .postplus/
    目录外
  • 在执行高成本的采集步骤前，将用户意图编译为小型Actor输入文件
  • 从有限的首次执行开始：
    • 一个URL
    • 一家店铺
    • 一个小型品类样本
  • 如果hosted功能不可用、未授权或返回稳定的网络错误，请立即停止，不要切换到临时shell脚本
如果用户未指定路径，请在
.postplus/
下为中间文件选择清晰的描述性名称，并明确最终输出路径。

Minimal Toolchain

最小工具链

Use these pieces in combination, not as a rigid pipeline:
  • collection runner: 
    • ${CLAUDE_SKILL_DIR}/scripts/collection_actor_run.mjs
  • normalization: 
    • ${CLAUDE_SKILL_DIR}/scripts/normalize_tiktok_shop_dataset.mjs
  • analysis: 
    • ${CLAUDE_SKILL_DIR}/scripts/analyze_tiktok_shop_dataset.mjs
The scripts are modular on purpose:
  • scrape only when the user wants data collection
  • normalize when downstream portability matters
  • analyze when the user wants decisions, not raw rows
Run the released collection through:
bash
node ${CLAUDE_SKILL_DIR}/scripts/collection_actor_run.mjs \
  --collection-key tiktok-shop-products \
  --input <work-folder>/.postplus/tiktok-shop/input.json \
  --output <work-folder>/.postplus/tiktok-shop/raw.json
组合使用以下组件，而非作为刚性流水线：
  • 采集运行器： 
    • ${CLAUDE_SKILL_DIR}/scripts/collection_actor_run.mjs
  • 标准化工具： 
    • ${CLAUDE_SKILL_DIR}/scripts/normalize_tiktok_shop_dataset.mjs
  • 分析工具： 
    • ${CLAUDE_SKILL_DIR}/scripts/analyze_tiktok_shop_dataset.mjs
这些脚本设计为模块化：
  • 仅当用户需要数据采集时执行抓取
  • 当下游可移植性重要时执行标准化
  • 仅当用户需要决策支持而非原始数据时执行分析
通过以下命令运行已发布的采集任务：
bash
node ${CLAUDE_SKILL_DIR}/scripts/collection_actor_run.mjs \
  --collection-key tiktok-shop-products \
  --input <work-folder>/.postplus/tiktok-shop/input.json \
  --output <work-folder>/.postplus/tiktok-shop/raw.json

Analysis Guidance

分析指南

The current analysis layer is best for:
  • price distribution
  • category frequency
  • dominant shops or brands
  • top products by orders or review volume
  • benchmark-candidate ranking
It is not a full business intelligence system.
If the user asks for something more specific, use the normalized dataset as the source of truth and compute the extra slice rather than rewriting the scrape layer.
当前分析层最适用于：
  • 价格分布
  • 品类频率
  • 主流店铺或品牌
  • 按订单量或评论量排序的热门商品
  • 标杆候选商品排名
它并非完整的商业智能系统。
如果用户提出更具体的需求，请以标准化数据集为事实来源，计算额外的细分数据，而非重写抓取层。

Failure and Fallback

失败与回退

If a run fails:
  1. check whether the page type matches 
    tiktok-shop-products
  2. reduce scope to a smaller sample
  3. preserve the raw response or error context
Do not silently switch collection keys.
如果运行失败：
  1. 检查页面类型是否匹配
    tiktok-shop-products
  2. 将范围缩小为更小的样本
  3. 保留原始响应或错误上下文
请勿静默切换collection key。

Things To Avoid

注意事项

Do not:
  • assume every request is a fixed "选品" workflow
  • hard-code collection fields into analysis logic
  • require the user to speak in collection input terms
  • bundle scraping, normalization, analysis, and reporting into one giant step by default
This skill should stay as a decision guide plus a small reusable toolkit.
请勿：
  • 假设每个请求都是固定的“选品”工作流
  • 将采集字段硬编码到分析逻辑中
  • 要求用户使用采集输入术语表达需求
  • 默认将抓取、标准化、分析和报告捆绑为一个大型步骤
本技能应作为决策指南及小型可复用工具集存在。

Files

文件列表

  • package.json
  • schemas/tiktok-shop-dataset.schema.json
  • scripts/lib/tiktok_shop_normalize.mjs
  • scripts/normalize_tiktok_shop_dataset.mjs
  • scripts/analyze_tiktok_shop_dataset.mjs
  • templates/pratikdani-product-urls.json
  • templates/jeremy-product-urls.json
  • package.json
  • schemas/tiktok-shop-dataset.schema.json
  • scripts/lib/tiktok_shop_normalize.mjs
  • scripts/normalize_tiktok_shop_dataset.mjs
  • scripts/analyze_tiktok_shop_dataset.mjs
  • templates/pratikdani-product-urls.json
  • templates/jeremy-product-urls.json