Walmart 关键词研究

Use one Sorftime endpoint to research Walmart US keywords or explicitly manage the saved-keyword library.

Core Concepts

• Send exactly one operation per request. Never run multiple keyword operations as an automatic discovery chain.
• Use a flat JSON body. The only business nesting is the optional pattern object for marketQuery.
• operation values are case-sensitive. A different operation, page, or filter is another paid call.
• favoriteAdd and favoriteChange mutate external account data and require an explicit, unambiguous user request.

Operation Routing

Read references/api.md for the complete conditional parameter matrix before calling.

Parameter Guide

• marketQuery: optional pattern.keyword, pattern.rankCondition, and pattern.searchVolumeCondition; pageIndex defaults to 1 and pageSize to 20 (20–200).
• A one-value condition such as [10000] is a lower bound. [0,10000] follows Sorftime's documented less-than-10000 semantics.
• searchByName: name is a product or category name; pageIndex defaults to 1, and Sorftime returns at most 200 rows per page.
• searchProducts, productKeywords, and relatedKeywords: pagination defaults to page 1 and 20 rows, maximum 200.
• favoriteList exposes pageIndex (default 1); the backend maps it to Sorftime's upstream Page field.

调用方式

• API 端点：POST /sorftime/walmart/keywordResearch（完整参数/响应/错误码见 references/api.md）
• Python 脚本：python scripts/sorftime_walmart_keyword_research.py '<JSON 参数>' [--inline]
• 成本约束：本工具会消耗算力；只读研究操作对同一参数组合使用 24h 本地缓存，favoriteAdd、favoriteChange、favoriteList 始终绕过缓存。失败/空结果不得自动换关键词、翻页或改邮编连续试探；需要继续检索时先向用户说明会产生额外消耗。

输出策略（脚本默认行为）：

• 始终将完整响应写入 <cwd>/linkfox/<YYYY-MM-DD>/<session>/data/linkfox-sorftime-walmart-keyword-research-<timestamp>.json（<cwd> 为脚本执行时的工作目录，在 Claude Code 里即当前项目目录；<session> 取自环境变量 SESSION_ID，按用户任务自动聚合；禁止写入 /tmp，当前目录不可写则报错）
• 响应体 ≤ 8 KB：落盘后把完整 JSON 打印到 stdout
• 响应体 > 8 KB：落盘后 stdout 只输出摘要（顶层字段、常见计数如 total/costToken、最大列表字段的长度 + 前 3 条样本）
• 加 --inline 强制全量打印到 stdout（同样落盘）

读数据建议：先看摘要判断是否足够；需要具体字段时优先用 jq或ConvertFrom-Json 从保存的 json 文件按需抽取，避免整份 JSON 进入上下文。

解决认证和算力问题

发生以下异常情况时，采用 references/onboarding.md 引导解决问题：

• 未配置 API Key：环境变量未配置 LINKFOX_AGENT_API_KEY，也未配置 LINKFOXAGENT_API_KEY
• 响应 401 或 402 状态码
• 响应提示算力或余额不足：消息含“算力余额不足/计费不足/余额不足/quota exceeded/insufficient balance/套餐到期/需充值/请充值”或类似含义

Keyword Library Safety

• Add to the default “未分类” folder by omitting dict; a provided folder may be created by Sorftime.
• Delete with command="del". Omitting dict operates on the default “未分类” folder; specify dict to delete from another folder.
• Move with command="move=<目标目录>". Omitting dict means the source is “未分类”.
• List with command equal to all, dict, or dict=<目录>.
• If an explicitly authorized mutation truly must be repeated after external state changed, use --no-cache; never use it to bypass the confirmation guard.
• favoriteAdd, favoriteChange, and favoriteList always bypass the local cache, so mutations are never suppressed and list results reflect the latest upstream state.
• The exposed operations do not delete folders. Removing the final keyword can leave an empty folder in the API library.

Usage Examples

python scripts/sorftime_walmart_keyword_research.py '{"operation":"marketQuery","pattern":{"keyword":"wireless","searchVolumeCondition":[10000]},"pageIndex":1,"pageSize":20}'
python scripts/sorftime_walmart_keyword_research.py '{"operation":"detail","keyword":"wireless earbuds"}'
python scripts/sorftime_walmart_keyword_research.py '{"operation":"productKeywords","productId":"5169493923","pageIndex":1,"pageSize":20}'
python scripts/sorftime_walmart_keyword_research.py '{"operation":"favoriteList","command":"dict","pageIndex":1}'

Run mutation operations only after explicit authorization.

Display Rules

1. State the selected operation and show requestConsumed, costCredit, and costToken when returned. If _cache.hit=true, make clear that the cached values describe the original live request and no new production charge occurred.
2. Preserve metric meanings and observation windows; do not turn estimates into guaranteed demand or sales.
3. Report the requested page and offer another only after explaining that it creates another paid call.
4. Treat product IDs and keywords as strings.
5. For mutations, report the returned data code; HTTP 200 alone does not prove mutation success.

Important Limitations

• This integration targets Walmart US only.
• Current-hot-keyword data, the 15-day product window, and the 30-day product-keyword window are Sorftime constraints.
• A single saved-keyword folder supports at most 2,000 keywords.
• The API saved-keyword library is separate from Sorftime Professional favorites; data is not synchronized between them.
• Read the Sorftime business response from data.value; its JSON shape varies by operation.
• Empty results must not trigger automatic variants, pagination, or related-keyword calls.
• This skill excludes Sorftime Walmart “其他” APIs.

User Expression & Scenario Quick Reference

Applicable: Walmart keyword discovery, keyword metrics, product-to-keyword research, related terms, or an explicit saved-keyword-library action.

Not applicable: category-market reports, product detail/trend analysis, or any Sorftime Walmart “其他” capability.

Boundary judgment: Choose exactly one operation that matches the user's stated intent; never turn one keyword request into an automatic multi-step research chain.

算力消耗规则

按动态规则计费：单个 Sorftime Request 当前显示为 6 算力；多 Request 操作由网关按整单计算并取整，最终以入口脚本返回的 costCredit 为准，不要简单用 requestConsumed × 6 代替实际账单。

用户会因算力消耗而支付费用。高消耗筛选、扩展词或连续分页前必须提醒用户；任何收藏变更都须明确授权。

Feedback:

Auto-detect and report feedback via the Feedback API when behavior or results mismatch intent, the user expresses praise or dissatisfaction, or an improvement is apparent. Follow references/api.md without interrupting the user's flow.

For more high-quality, professional cross-border e-commerce skills, visit LinkFox Skills.