MCP server for Wildberries Seller API: 197 tools for products, prices, orders, ads, supplies, reviews, finance and analytics, with response sizes measured on a live account and trimmed from 770k to 75k tokens across a 27-response corpus
23 tools with explicit schemas and descriptions, but significant quality gaps. All tools have descriptions (10-200 chars, mostly 50-150), and input schemas are present with basic type definitions. However, parameter descriptions are sparse or missing entirely, most params lack explanations of what they control or accept. Output schemas are undocumented. Error handling is absent from descriptions. No guidance on idempotency, retryability, or recovery. Tool names follow verb_noun convention (wb_list_*, wb_card_*, wb_tag_*) but lack clarity on mutual exclusivity and dependencies (e.g., wb_cards_update vs wb_cards_create both modify state but differ in semantics). Schemas use 'object' types for complex params (cursor, filter, characteristics) without nested field documentation. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE). Composition is reasonable, tools chain via shop_id and nm_id, but output structure is not documented, forcing LLMs to guess what fields are returned.
Generate barcodes for new goods (баркоды).
Seller brands; subject_id filters by category (бренды).
Card details by nmID (карточка товара).
[P0] Cards with errors: blocked/rejected, incl. IP-holder complaints (заблокированные карточки). Check regularly.
Recommended goods on a card, cross-sell (рекомендации).
Set recommended goods on a card, max 10 nmID (рекомендации, средний чек).
Complex object parameters (cursor, filter, characteristics) lack nested field documentation. LLMs cannot construct valid payloads without knowing what fields these objects require.
No output schemas documented. Tool descriptions state what they do but not what fields are returned, forcing LLMs to guess response structure and breaking tool chaining.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite explicit risk levels (READ_ONLY, WRITE, DESTRUCTIVE). LLMs cannot determine which tools are safe to retry or have side effects.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
Create new cards, async, up to 30 min to sync. Get subject characteristics via wb_subject_charcs first (создать карточки).
Card create/edit limits (лимиты карточек).
List product cards (карточки товаров), cursor pagination.
Move cards to trash by nmID (удалить карточки).
Restore cards from trash by nmID (восстановить карточки).
Cards in trash (корзина).
Update cards: description, SEO, characteristics (обновить карточку).
WB reference books: colors, kinds, countries, seasons, vat, tnved (справочники).
Registered WB shops (магазины): shop_id + name. Use shop_id in all other tools.
Upload photo/video to a card by URL. Warning: REPLACES all existing media (медиа, фото).
Subject characteristics: required/optional card fields (характеристики предмета).
Search WB subjects/categories for card creation (предметы, категории).
Create a tag; color is HEX (создать ярлык).
Delete a tag. Fails if still linked to goods — unlink via wb_tag_link first (удалить ярлык).
Attach/detach tags on a card. Pass the FULL new tag list (привязать ярлык).
Update a tag: name, color (обновить ярлык).
Seller tags for grouping cards (ярлыки, теги).
Parameter descriptions missing or generic. 'filter' and 'cursor' have only type hints; 'characteristics' and 'variants' lack any explanation of required/optional fields or format.
No error handling guidance in descriptions. Tools like wb_cards_create (async, 30 min sync) and wb_media_upload (replaces all media) lack recovery instructions or warnings about side effects.