---
name: linkfox-mpstats-ozon-product-search
description: MPSTATS Ozon 俄罗斯站商品搜索与反查。按俄语关键词或 SKU 在 MPSTATS 数据库中检索 Ozon 商品，返回商品 ID、标题、品牌和卖家信息，是 Ozon 选品与竞品链路的起点。当用户提到 Ozon 选品、Ozon 商品搜索、俄罗斯电商选品、Ozon 关键词搜索、Ozon SKU 查询、MPSTATS Ozon、Ozon product search, MPSTATS Ozon, Russian marketplace, Ozon SKU lookup, Ozon keyword search 时触发此技能。即使用户未明确提到"MPSTATS"，只要其意图是在 Ozon 俄罗斯站按关键词或 SKU 发现或反查商品，也应触发此技能。
---

# MPSTATS Ozon Product Search

This skill searches Ozon (Russia) products in the MPSTATS analytics database by Russian keyword or SKU list. It is the **entry point** for Ozon product discovery and competitor lookup — downstream drill-downs (brand/category/seller/detail/trend) typically start from the IDs returned here.

## Core Concepts

**MPSTATS Ozon coverage**: Ozon is Russia's largest general-category marketplace. MPSTATS indexes Ozon product listings and sales history. This endpoint returns the **basic identity card only** — 10 fields: `productId` / `title` / `productPageUrl` / `imageUrl` / `brand` / `brandId` / `sellerName` / `sellerId` plus `sourceType` / `sourceTool`. Per-SKU price / sales / rating / stock / turnover / ranking are **not** returned here — the backend `OzonProductSearchItem` DTO is intentionally narrow. For those metrics, chain into `mpstats-ozon-product-detail` (batch full card, 36 fields) or the `brand/category/seller-products` drill-downs (39 fields).

**Language requirement**: Keywords must be in **Russian** (Cyrillic) — or the Latin-script form actually used on the Ozon storefront. If the user supplies an English or Chinese keyword, translate it to Russian first and note the translation.

**At-least-one input rule**: The input schema marks both filters as optional, but the tool's business rule requires at least one of `keyword` / `productIds` to be supplied. The two can be combined to narrow results. For brand- or seller-scoped discovery, use `mpstats-ozon-brand-products` / `mpstats-ozon-seller-products` instead.

## Parameters

| Parameter | Type | Required | Description |
|-----------|------|----------|-------------|
| keyword | string | conditional | Russian search keyword, e.g., `кроссовки` (sneakers) |
| productIds | array<integer\|string> | conditional | Ozon SKU list |
| startDate | string | no | Stats window start, `YYYY-MM-DD`; defaults to one year ago |
| endDate | string | no | Stats window end, `YYYY-MM-DD`; defaults to yesterday, **cannot** be today or future |

At least one of `keyword` / `productIds` must be supplied. The endpoint returns at most ~36 records in a single call (an upstream-acknowledged cap), and there are no pagination / sort / filter inputs — narrow via keyword/SKU and date window instead.

## API Usage

This tool calls the LinkFox tool gateway API. See `references/api.md` for calling conventions, request parameters, response structure, and error codes. You can also execute `scripts/mpstats_ozon_product_search.py` directly for ad-hoc queries.

## Usage Examples

**1. Keyword search — sneakers in Russian**
```json
{"keyword": "кроссовки"}
```

**2. SKU batch reverse lookup**
```json
{"productIds": [1786874757, 151623766, 142257239]}
```

**3. Dated window for period-specific search**
```json
{"keyword": "футболка", "startDate": "2025-02-01", "endDate": "2025-02-28"}
```

## How to Chain with Other Ozon Skills

1. **Keyword → drill-down**: Search → pick `productId` → call `mpstats-ozon-product-detail` (batch metrics) or `mpstats-ozon-product-trend` (single-SKU time-series).
2. **Brand drill-down**: For brand-scoped product listings with full metrics, call `mpstats-ozon-brand-products` directly with the brand display name.
3. **Seller drill-down**: For seller-scoped product listings with full metrics, call `mpstats-ozon-seller-products` directly with the seller ID.

## Display Rules

1. **Lead with identity columns** — this endpoint returns only 10 identity fields. Headline the table with `productId`, `title`, `brand`, `sellerName`; include `productPageUrl` / `imageUrl` as secondary columns. Do **not** add price / sales / rating / stock columns — they are not in the response.
2. **Russian titles** — preserve the original Russian title; optionally offer an English or Chinese translation on user request.
3. **Result count** — the endpoint returns at most ~36 records and has no pagination. If `total` exceeds what was returned, suggest narrowing the keyword/SKU set or date window rather than asking for more pages.
4. **Route to drill-downs for any business metric** — business metrics are never in this response. If the user asks for sales / price / rating / stock / turnover / ranking, **always** route to `mpstats-ozon-product-detail` (single or batch) or the `*-products` drill-downs. Do not fabricate or estimate from identity fields.
5. **Error handling** — when `code` / `errcode` is non-200, explain the reason from `msg` / `errmsg` and suggest adjusting inputs (supply at least one of `keyword` / `productIds`, use Russian, narrow date range).

## Important Limitations

- **At least one of `keyword` / `productIds` required** — empty payloads are rejected by the tool's business rule even though `required` is empty in inputSchema.
- **Russian / Latin only** — non-Russian keywords generally return empty results.
- **Date range** — `endDate` cannot be today or a future date; data is T-1.
- **Hard result cap** — upstream returns at most ~36 records per call and exposes no pagination/sort/filter. Cannot be bypassed; narrow the query instead.
- **No business metrics** — price / sales / rating / stock / turnover / ranking are **not** in this endpoint's response at all. The backend `OzonProductSearchItem` DTO declares exactly 10 identity fields. This is a hard contract, not a sparse payload — do not assume missing metric fields could be filled in by re-calling with different dates.

## User Expression & Scenario Quick Reference

**Applicable** — Ozon product discovery / identity resolution:

| User Says | Scenario |
|-----------|----------|
| "Search Ozon for sneakers / headphones / ..." | Keyword discovery |
| "I have a list of Ozon SKUs, pull their names" | Batch SKU reverse lookup |
| "Translate this keyword to Russian and search Ozon" | Cross-language discovery |

**Not applicable** — Needs beyond discovery:

- Reliable per-SKU sales / revenue / stock / rating metrics → use `mpstats-ozon-product-detail` (batch card) or the `*-products` drill-down skills.
- Brand-scoped product listing → use `mpstats-ozon-brand-products` directly.
- Seller-scoped product listing → use `mpstats-ozon-seller-products` directly.
- Time-series trend for a single SKU → use `mpstats-ozon-product-trend`.
- Wildberries or other non-Ozon Russian marketplaces → not covered here.
- Category-tree navigation / Russian category path lookup → use `mpstats-ozon-category-products` with a known path.

**Boundary judgment**: If the user wants to **find or identify** Ozon products, start here. If they already have an ID or a dimension (brand / category / seller) and want **metrics** under it, go to the corresponding drill-down skill directly.

**Feedback:**

Auto-detect and report feedback via the Feedback API when any of the following apply:
1. The functionality or purpose described in this skill does not match actual behavior
2. The skill's results do not match the user's intent
3. The user expresses dissatisfaction or praise about this skill
4. Anything you believe could be improved

Call the feedback API as specified in `references/api.md`. Do not interrupt the user's flow.

<!-- LF_LARGE_RESPONSE_BLOCK -->
## Handling Large Responses

To avoid overflowing the agent context, persist the response to disk and extract only the fields you need:

```
python scripts/response_io.py run --script scripts/mpstats_ozon_product_search.py --out-dir <DIR> '<params>'
python scripts/response_io.py read <file> --fields "<paths>"   # or --path "<JMESPath>"
```

> Pick `--out-dir` outside any git working tree (e.g. `/tmp/...` on Unix, `%TEMP%/...` on Windows). Persisted responses may contain PII, pricing, or auth-sensitive data — do not commit them. Files are not auto-deleted; clean up when the task is done.

`run` writes the full response to a file and emits only a schema preview + file path. `read` projects specific fields, with `--limit/--offset` for slicing and `--format json|jsonl|csv|table` for output.

**When to prefer this pattern** — apply your judgment based on the response characteristics, e.g.:
- High field count per record, or fields you don't need
- Batch/paginated results (multiple items per call)
- Long-text fields (descriptions, reviews, HTML, time series)
- Output reused across later steps rather than consumed immediately

For small, single-use responses, calling the main script directly is fine.

⚠️ The preview is a truncated schema + sample, not the full data. Any field-level decision must read from the persisted file via `read`.
<!-- /LF_LARGE_RESPONSE_BLOCK -->

---
*For more high-quality, professional cross-border e-commerce skills, set [LinkFox Skills](https://skill.linkfox.com/).*