Skip to main content
Glama
812,289 tools. Updated 2026-11-20 18:15

"A product search on eBay for a specific item" matching MCP tools:

  • Re-send the LIVE eBay listing of up to 25 products from each product as it is now — title, description, photos, item specifics, condition, price, quantity and business policies — without ending the sale. Use after a listed product was corrected (update_product, upsert_product_translation, upsert_product_attributes, upsert_product_prices, set_channel_texts) and eBay should show it; a draft goes through publish_ebay_drafts instead, and the listing's eBay category never changes here (recreate_ebay_drafts does that). Every go-live gate runs again, so a missing required specific is named rather than left for eBay; a product whose eBay sync is paused is refused. Needs the owner's AI publishing switch and an armed listingCreate direction. eBay allows 250 revisions per listing a day. Returns done, failed (named, with reasons) and notAttempted. The result echoes the workspace.
    Connector
    Destructive
    No auth
  • Returns normalized details for a public eBay item listing. item_id: eBay listing item ID, e.g. 196435705622. Use ebay_search to search eBay listings.
    ConnectorNo auth
  • Returns normalized details for a public eBay item listing. item_id: eBay listing item ID, e.g. 196435705622. Use ebay_search to search eBay listings.
    ConnectorNo auth
  • Search and browse product listings across multiple eBay international sites. Trigger when user mentions eBay product search, eBay listing query, eBay price comparison, eBay market browsing, eBay sold items, eBay auction search, eBay product research, eBay products, eBay pricing, eBay competitors, eBay auctions, eBay market analysis — even if "eBay" is not explicitly mentioned, as long as the need involves searching products on eBay, comparing eBay prices, finding sold listings, or analyzing eBay market data.
    ConnectorOAuth
  • Search and browse product listings across multiple eBay international sites. Trigger when user mentions eBay product search, eBay listing query, eBay price comparison, eBay market browsing, eBay sold items, eBay auction search, eBay product research, eBay products, eBay pricing, eBay competitors, eBay auctions, eBay market analysis — even if "eBay" is not explicitly mentioned, as long as the need involves searching products on eBay, comparing eBay prices, finding sold listings, or analyzing eBay market data.
    ConnectorOAuth
  • Remove a specific image from a product. Destructive, idempotent. Authenticated. OAuth (scope `products:write`) preferred; `api_key` fallback. Use when an image was uploaded by mistake or the merchant updated their listing. The product itself is preserved — only the image record and its file are removed. To remove the product entirely use `delete_product`. Args: product_id: ID of the product the image belongs to. image_id: ID of the image to delete. Visible in the `images` array of `get_product` responses. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: ``{"deleted": True, "product_id": int, "image_id": int}`` on success, or ``{"error": ...}`` on auth/ownership failure.
    Connector
    Destructive
    No auth

Matching MCP Servers

  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI agents to call tools through a security gateway that intercepts requests, enforces deterministic policy and risk checks, blocks or escalates high-risk actions for human approval, and keeps tool credentials and audit logs protected.
    1
    MIT

Matching MCP Connectors

  • Git AOAuth

    Many AI agents on one repo, every change lands on main. Read code, propose changes, review, read CI.

  • Indicative UK skip hire prices by town, council skip permit rules and skip size recommendations.

  • Remove a specific image from a product. Destructive, idempotent. Authenticated. OAuth (scope `products:write`) preferred; `api_key` fallback. Use when an image was uploaded by mistake or the merchant updated their listing. The product itself is preserved — only the image record and its file are removed. To remove the product entirely use `delete_product`. Args: product_id: ID of the product the image belongs to. image_id: ID of the image to delete. Visible in the `images` array of `get_product` responses. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: ``{"deleted": True, "product_id": int, "image_id": int}`` on success, or ``{"error": ...}`` on auth/ownership failure.
    Connector
    Destructive
    No auth
  • Search for products on Kapruka.com by keyword, with optional category filter and pagination. Returns a ranked list of matching products with prices, stock status, images, and URLs, plus the category facets available for the query (facets.categories) — the names that work in `category` to narrow the same search. Supports cursor-based pagination — pass next_cursor from one response into the next call. Pagination is capped at 3 pages per query to discourage catalog enumeration; for broader discovery, refine the query or narrow with a facet instead. Queries must be at least 3 characters and contain specific terms — pure stopword queries (e.g. "the", "a an") are rejected. Relevance: the API matches ANY single query word, so extra words add loosely related results rather than narrowing. Search the thing the customer wants (the head noun) and check the product names actually contain it before presenting them. Prices and min_price/max_price are both in `currency` — pass the customer's own budget in their own currency; do not convert it. Gifts for recipients OUTSIDE Sri Lanka are not in this search: use kapruka_gift_countries / kapruka_gift_country_products. A row's `price` is the discounted price checkout charges (the bounds filter on it); `compare_at_price` is set when on sale. Search results carry NO delivery-scope information. Food, hotel cakes and liquor are delivered only to selected cities — never infer deliverability from a search hit. Before quoting delivery on a specific item, call kapruka_get_product (read `delivery.island_wide`) or kapruka_check_delivery with `product_id`. Args: params (SearchProductsInput): - q (str): Search query (e.g. 'birthday cake', 'roses', 'tea gift'). Min 3 chars. - category (Optional[str]): A facet name from a previous search's facets.categories - limit (int): Results per page, 1–50 (default 10) - cursor (Optional[str]): Pagination cursor from previous response - currency (str): LKR (default), USD, GBP, AUD, EUR - min_price (Optional[float]): Min price (inclusive) in `currency` - max_price (Optional[float]): Max price (inclusive) in `currency` - in_stock_only (bool): Restrict to in-stock items (default false) - sort (str): 'relevance' | 'price_asc' | 'price_desc' | 'newest' | 'bestseller' - include_stubs (bool): Deprecated, no effect - response_format (str): 'markdown' (default) or 'json' Returns: str: Search results in the requested format. JSON schema: { "results": [ { "id": str, "type": "product", "name": str, "summary": str, "price": {"amount": float | null, "currency": str}, "compare_at_price": {"amount": float, "currency": str} | null, "in_stock": bool, "stock_level": str, "image_url": str | null, "category": {"id": str, "name": str, "slug": str}, "rating": null, "ships_internationally": bool, "url": str } ], "next_cursor": str | null, # null after page 3 even if upstream has more "total_estimate": int, # index hits for the query words (any-word match) "applied_filters": {"q": str, "category": str, "currency": str, ...}, "facets": {"categories": [{"name": str, "count": int}]}, "category_filter_dropped": str, # only when `category` was not a facet here "valid_categories": [str] # ...and the facet names that are } Error: "Error: <message>" or "No products found for '<query>'" on failure.
    ConnectorNo auth
  • ADVANCED / manual selection. Do NOT use this to build a shopping list: for any list of 2 or more items, call build_basket (one call, server-side, cheapest pick per item). Use batch_search only when you need the RAW candidate products for items you intend to select MANUALLY (for example, to override build_basket's pick on one item). It searches multiple items in parallel and returns raw candidate products per query (up to candidates_per_retailer per retailer) with NO filtering or scoring applied. WORKFLOW: 1. Call batch_search with your shopping list items 2. Review the candidates for each item. For each query, pick the single best product_id per retailer that actually matches what the user wants (correct product type, reasonable size, not a different product). If none of the candidates are a good match for a query, call batch_search again with a refined query (e.g., more specific terms, different wording) 3. Once you have confirmed product_id selections for all items, call save_basket with your picks TIPS: - Set product_type to help narrow results to the right category (e.g., fresh_produce for herbs and vegetables) - Each candidate includes a product_categories field (JSON array of retailer category labels like 'Yoghurt', 'Herbs & Spices', 'Sausages'). Use this to distinguish the actual product type when names are ambiguous (e.g., 'oregano' the herb vs 'oregano sausages') - Prefer products where the name clearly matches the query intent - For staples (milk, eggs, bread), the cheapest option is usually the store brand - If results look wrong (e.g., searching 'rosemary' returns snacks instead of herbs), retry with a more specific query like 'fresh rosemary' - Each item includes a 'cheapest' hint with one entry per retailer. This is the cheapest candidate whose product_categories match the inferred subcategory for the query, NOT necessarily the lowest unit_price overall: a clearly off-category cheaper item (e.g., a snack bar under 'peanut butter') is deliberately excluded. When no category signal is available it falls back to the raw lowest unit_price. Treat the hint as advisory and confirm against the full candidate list and the user's intended format. AMBIGUITY: When candidates for a query span clearly different product formats (e.g., microwave rice pouches vs dry rice bags, single-serve vs family/bulk sizes, flavoured vs plain variants, fresh vs frozen), ASK the user which format they prefer before selecting a product. Compare the size field across candidates to detect this: a spread from 125g to 5kg signals format ambiguity. Do not silently pick the cheapest option when it may be the wrong format (e.g., a 250g microwave pouch when the user likely wants a 1kg bag).
    ConnectorNo auth
  • ADVANCED / manual selection. Do NOT use this to build a shopping list: for any list of 2 or more items, call build_basket (one call, server-side, cheapest pick per item). Use batch_search only when you need the RAW candidate products for items you intend to select MANUALLY (for example, to override build_basket's pick on one item). It searches multiple items in parallel and returns raw candidate products per query (up to candidates_per_retailer per retailer) with NO filtering or scoring applied. WORKFLOW: 1. Call batch_search with your shopping list items 2. Review the candidates for each item. For each query, pick the single best product_id per retailer that actually matches what the user wants (correct product type, reasonable size, not a different product). If none of the candidates are a good match for a query, call batch_search again with a refined query (e.g., more specific terms, different wording) 3. Once you have confirmed product_id selections for all items, call save_basket with your picks TIPS: - Set product_type to help narrow results to the right category (e.g., fresh_produce for herbs and vegetables) - Each candidate includes a product_categories field (JSON array of retailer category labels like 'Yoghurt', 'Herbs & Spices', 'Sausages'). Use this to distinguish the actual product type when names are ambiguous (e.g., 'oregano' the herb vs 'oregano sausages') - Prefer products where the name clearly matches the query intent - For staples (milk, eggs, bread), the cheapest option is usually the store brand - If results look wrong (e.g., searching 'rosemary' returns snacks instead of herbs), retry with a more specific query like 'fresh rosemary' - Each item includes a 'cheapest' hint with one entry per retailer. This is the cheapest candidate whose product_categories match the inferred subcategory for the query, NOT necessarily the lowest unit_price overall: a clearly off-category cheaper item (e.g., a snack bar under 'peanut butter') is deliberately excluded. When no category signal is available it falls back to the raw lowest unit_price. Treat the hint as advisory and confirm against the full candidate list and the user's intended format. AMBIGUITY: When candidates for a query span clearly different product formats (e.g., microwave rice pouches vs dry rice bags, single-serve vs family/bulk sizes, flavoured vs plain variants, fresh vs frozen), ASK the user which format they prefer before selecting a product. Compare the size field across candidates to detect this: a spread from 125g to 5kg signals format ambiguity. Do not silently pick the cheapest option when it may be the wrong format (e.g., a 250g microwave pouch when the user likely wants a 1kg bag).
    ConnectorNo auth
  • Find and buy products, tickets, bookings and subscriptions for the user at supported stores. Start with `search` { query } to find the item and merchant page using Vaaya OneSearch (5¢ per search); the user can describe what they need without supplying a link. Before checkout, use `setup` to check payment and shipping readiness; resolve missing setup once. Persistent merchant accounts are managed at /store-connections. Use store_connection_id to select an account. Passwords and verification codes belong only on the merchant page, never in chat. PRODUCT DISCOVERY: if the user describes what to buy but provides no URL, use `search` { query: <product, variant, budget and preferred merchant if specified> } to locate it under the existing tool spending permissions; do not ask them to paste a link or say search for it first. Search is not purchase authorization. Respect search fees and spending limits; if search is unavailable, explain the actual error and then ask for a link. Verify the merchant, item and variant from the result; ask only when matches are ambiguous or details cannot be verified. Never invent a product URL, final total or delivery date from a search snippet. THE SEAMLESS PATH: once the user has told you what to buy and you have the exact item, merchant page URL and total, call `purchase` { item, merchant, url, total_cents, confirmed: true, confirmation: <the user's own words> } — it approves from their message (under the chat limit), starts buying in the user's cloud browser in the background and returns `message` ("Hold on — buying it now."): RELAY IT, then poll `status` { approval_id } every ~10 seconds and relay its `message` when the status is completed ("Done — …"), requires_action (read action_required.reason for the exact blocker) or failed. If `status` says the user's shipping address is missing, ask for it and call `address` { name, line1, line2?, city, state?, postal_code, country, phone? } then `checkout` { approval_id } to resume; if merchant authentication is pending, relay handoff_url and handoff_deadline. The durable worker checks authentication and resumes the same authorized purchase automatically. Do not ask the user to send done or submit codes in chat. If the handoff expires, relay the recovery page; it does not clear order or payment uncertainty. Other sub-commands: `search` { query } (protocol merchants plus a real web search; the web half bills 5¢), `propose` { item, merchant, total_cents, url?, notes?, confirmed?, confirmation? } (creates an approval; without `confirmed` it returns an approval link the user opens — show its `message` VERBATIM), `checkout` { approval_id, params? } (buys an approved purchase; for a browser merchant it runs in the background like `purchase`). Use the user's existing authorization of the exact item and total; do not ask them to confirm twice. Ask only for missing purchase details. Nothing is ever bought without the user's yes: `checkout` refuses anything else. Use `reconcile` { approval_id } after an uncertain submission: it only inspects the existing checkout and never submits payment. For an unavailable checkout or inconclusive reconciliation, relay recovery_url and recovery_instructions from status. The user can resolve the attempt on its approval page only after checking both merchant orders and payment records. A chat statement alone does not clear the lock, and expiration does not clear it. If status or replay says purchase_resolved, never reuse checkout on that approval. When the user requests another attempt, call propose with retry_of set to that resolved ID, current item/price and no confirmed flag, then show the new approval link. Check setup browser_allowance first; a daily_browser_limit requires waiting until resets_at, not reconnecting or changing request text. For a handoff, return action_required.url and the deadline. The user fixes 2FA, CAPTCHA, payment, address or booking issues on that authenticated page and uses Return control to Vaaya; do not call checkout to bypass a pending handoff. Status polling never creates a replacement. Resume only the same approval_id; do not recreate a purchase to bypass an unresolved attempt. `charged_cents` is the Vaaya tool fee, NOT a merchant charge; read merchant_payment separately, and never infer a hold or capture from Link approval. Never open `browserbase` sessions yourself to buy something — only `buy` can.
    ConnectorNo auth
  • Change an item or record: changes holds only the fields to change. Fields are passed through as given and checked by the product, so read the item first and send the field names it reports; an unknown or malformed field comes back as an error naming it. Use move to change a folder, publish/unpublish to change live state. By type: • form — title, description, custom_slug, translation_id. It is live for respondents. Confirm with the user before calling this. Fields, pages, logic, emails, PDFs, submission settings and theme change with edit_source; anything neither covers (integrations, payments setup, sharing) is changed in the Paperform editor (fetch returns its url). • workflow — only a rename: { name, v, version_id }, with v and version_id from its #source's metadata. Steps, triggers and rules change with edit_source on its #source, never here (a schema is refused). Writes the draft only; nothing changes live until publish. • product, coupon — price and stock changes apply from the next submission; say so to the user first. • form_webhook, document_webhook — repointing sends future data somewhere new. Confirm with the user before calling this.
    ConnectorOAuth
  • Change one of the merchant's draft orders, as the order page does: its shipping method, 48H fulfillment on or off, its reference_id, or any of them. Its address and line items have tools of their own. 48H (made and shipped within 2 business days, for an added fee) needs at least one item on the order that offers it, and changes which methods the order can use. A method the order's address and items don't ship with is refused, naming the ones they do. reference_id is the merchant's own id for the order: one another of their orders has, unless it was cancelled, is refused, naming that order. A submitted or rejected order needs move_order_to_draft first. Returns the methods the order ships with now, none until every line item has a product, and its new total.
    Connector
    Destructive
    API key
  • Manage the TracePass product catalogue. A product is the catalogue layer — one product can have many passports (one per serialised unit). Products are not billable on their own. Archive is in tracepass_passport_lifecycle. Actions (pass via `action`; `{…}` is the `args` object): - list — { page?, limit? (≤100), category?, status?, search? }. Read-only. - get — { id }. Read-only. - create — { name, model, category, description? }. `category` is one of the 13 DPP categories (tracepass_templates list). - update — { id, name?, model?, description? }; at least one field. - get_links — { id, format? (json|csv), limit? (≤5000), cursor? (json: prior nextCursor) }. Every PUBLISHED passport's serial, passportUrl, status (json adds qrApiUrl, needs your API credentials) plus the model page URL, for a dealer asking for passport links. - get_model_qr — { id, format? (svg|png), gtin? (14 digits) }. QR of the model page. 404 model_not_found; 400 gtin_required (several GTINs) or invalid_gtin. - create_batch — { products: [ { name, model, category, description? }, … ] }, up to 100. Partial-success: per-item status, so some can be created while others error. The batch consumes N writes upfront; if that would exceed the daily cap NOTHING is created (429).
    Connector
    Destructive
    No auth
  • Search Partle's product catalog by name or description. CRITICAL SEARCH INSTRUCTION: Reason from the job to the product class first, then search with a descriptive product phrase (e.g. including substrate, material, or size class). DO NOT blindly search using the user's raw conversational words. Transform questions like 'what do I need to attach a mirror to a brick wall?' into a product phrase like 'heavy duty masonry wall anchor'. Two distinct modes: - **Default (no flags)** — fast keyword search. ~100ms. Acts like a normal "dumb" search box: matches the literal words you typed against product names and descriptions, with stemming. Good for queries where the user knows the product's likely name ("BC547", "Arduino Uno", "Bosch drill"). Returns noisy/wrong results on cross-language or attribute queries ("compost bin" matches Spanish "composta", not real composters). - **`super_search=True`** — slow, high-quality. ~1–2s. Run when the user describes what they want rather than naming it: cross-language ("Schraubenzieher Set" → real screwdriver sets even without German catalog entries), attribute-style ("small metal part with a flat head"), or any case where the default returns junk. Embeds the query with voyage-3-large, takes the cosine top-50 over the corpus (with an exact-name precision boost for part numbers), then a cross-encoder reranks them. The two modes are mutually exclusive in practice — pick one based on whether the user knows the product's name or is describing it. Use this when the user asks to find a specific product or browse products matching a query. Prefer over `search_stores` when the intent is product-led ("find a drill") rather than store-led. Use `get_product` afterwards if the user wants full details for one specific result. Read-only. No authentication. Rate-limited to 100 requests/hour per IP. Args: query: Free-text search term. In default mode, treated as keywords (each word matched against product text). In `super_search=True`, treated as a natural-language description. min_price: Lower bound on price in EUR. Omit for no lower bound. Null-priced rows are NOT excluded by this filter — pass `has_price=True` if you need only priced listings. max_price: Upper bound on price in EUR. Omit for no upper bound. Tip — narrow by budget: `min_price=10, max_price=50, sort_by="price_asc", has_price=True`. Products without a listed price (a large fraction of the scraped catalog) sort last under either price ordering and are kept in results unless `has_price` filters them out. tags: Comma-separated tag filter (e.g. "electronics,bluetooth"). Tags are AND-ed together. store_id: Restrict results to a single store. Use the integer `id` from `search_stores` results. sort_by: One of `price_asc`, `price_desc`, `name_asc`, `newest`, `oldest`. Omit to use the default search-relevance ranking. has_price: When True, exclude products without a listed price (~most of the scraped catalog). Use this for competitive pricing or budget-bounded shopping. When False, return only null-priced listings (rarely useful). Omit to include both. semantic: Legacy flag. Pure vector ordering, ~250ms. Mostly superseded by `super_search=True` (which uses the same vector retrieval plus a cross-encoder rerank for materially better ordering at the cost of another ~700ms). Keep using it only if you specifically want vector retrieval *without* the rerank. super_search: **Enable for natural-language / "describe what I want" queries.** ~1–2s. Embeds the query with voyage-3-large, takes the cosine top-50 (with a precision boost for exact-name matches like part numbers / SKUs), then a cross-encoder reranks them. Use whenever the user is describing rather than naming — cross-language ("Schraubenzieher Set"), attribute-style ("small black metal bracket"), or any case where the default keyword path returns junk. Don't combine with cheap browse-style queries where the user typed an exact product name — keyword default is faster there. On `relevance_score` here: better than the bi-encoder cosine, but still not a "did I find what the user wanted" gauge. Behavior to expect: gibberish or fully-off-topic queries cap around 0.35; loosely-related catalogue clusters can score 0.7+ even when no item truly matches (a "ceramic vase" query in a catalog with no vases but many ceramic flowerpots will still score high). **Read the product names** before claiming a match. The score is most useful as a relative signal within one result set — a sharp drop between rank N and N+1 marks where the catalog stops being useful for this query. limit: Max results (1–100, default 20). Larger limits are slower and consume rate budget faster. offset: Skip this many results before returning. Use for pagination (offset += limit on each follow-up call). Returns: A list of products. Each includes `id`, `name`, `price`, `currency`, `url`, `description`, `store` (id/name/address), `tags`, `images`, a canonical `partle_url`, and `relevance_score` (cosine similarity 0–1 between the query and the product's embedding when a query was provided; `None` otherwise). **Always share `partle_url` with the user so they can view the listing.** Caveat on `relevance_score`: it is monotonic *within a single search result set* (useful for spotting a big drop-off between rank 3 and rank 4), but its absolute value is not well-calibrated across queries — most results land in 0.55–0.80 regardless of whether the catalog has truly relevant items. Don't infer "this is a great match" from a 0.75 score alone.
    ConnectorNo auth
  • Create a new product listing on Partle. Authenticated. Prefer **OAuth**: connect once via the consent flow on claude.ai (or any MCP client that supports OAuth) and the bearer token is attached automatically — no `api_key` parameter needed. **Fallback**: pass an `api_key` (prefix `pk_`, generate at /account) for programmatic or non-OAuth clients. Required OAuth scope: `products:write`. Use when the user wants to add an item for sale. For edits to an existing product, use `update_product` instead. **Images.** This tool creates text fields only — no image arg. Do **not** try to pass image bytes through a tool argument; phone-sized payloads blow past conversation context limits. The response includes a one-shot ``upload_url`` (signed, ~15 min TTL, bound to this product and your authenticated user). To attach an image from your code-execution sandbox, do **one** PUT request — no auth headers needed, the URL itself carries the credential: requests.put(result["upload_url"], data=open("/path/to/photo.jpg", "rb").read(), headers={"Content-Type": "image/jpeg"}) The bytes flow Python → HTTP body → Partle, never through the conversation. The URL works once and expires fast. Alternative if you don't have local bytes but have a public image URL: call ``upload_product_image(product_id, image_url=...)`` instead. **Duplicate prevention.** Same user, same product name (case- and whitespace-insensitive) returns 409 with `existing.id`, `existing.url`, **and a fresh `upload_url`** for that existing product — so if the user is just retrying with a photo, you can attach it directly to the existing listing without having to create or pick anything new. You can also call `update_product` to change fields. Don't retry blindly. **Idempotency.** Pass `idempotency_key` (any unique string per logical create — UUID or hash of the source listing) and a retry after a network failure returns the original response instead of creating a duplicate. Reusing a key with a different payload is a 422. Args: name: Product name. Required, 1–200 chars. description: Long-form product description. Optional. price: Price in whole currency units, **not** cents (e.g. ``15.99`` means €15.99). Max 100000. Omit for "ask the seller". currency: Currency symbol. Defaults to `€`. Use `$`, `£`, etc. url: Link to the merchant's product page. Optional but recommended. store_id: ID of the store this product belongs to. Omit for a personal listing not tied to any store. listing_type: ``in_stock`` (default) when the seller has the item and it can be bought now. ``tentative`` when they do not stock it and want to measure interest first — such a listing is kept out of normal search results and instead collects "I need this" presses. Only use ``tentative`` if the user explicitly said they are gauging demand; an item that is merely out of stock today is still ``in_stock``. If the user is looking to *buy* something nobody sells, use `create_buy_request` instead — that is the demand side and it is a different tool. idempotency_key: Optional retry-safety token. Unique per logical create. Send the same key on retries to get the same response. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: The created product record including its new `id` and canonical `partle_url`. Share `partle_url` with the user. Returns ``{"error": ...}`` on auth, dedup, or validation failure (dedup also returns ``{"existing": {"id", "name", "url"}}``).
    ConnectorNo auth
  • Search Partle's product catalog by name or description. CRITICAL SEARCH INSTRUCTION: Reason from the job to the product class first, then search with a descriptive product phrase (e.g. including substrate, material, or size class). DO NOT blindly search using the user's raw conversational words. Transform questions like 'what do I need to attach a mirror to a brick wall?' into a product phrase like 'heavy duty masonry wall anchor'. Two distinct modes: - **Default (no flags)** — fast keyword search. ~100ms. Acts like a normal "dumb" search box: matches the literal words you typed against product names and descriptions, with stemming. Good for queries where the user knows the product's likely name ("BC547", "Arduino Uno", "Bosch drill"). Returns noisy/wrong results on cross-language or attribute queries ("compost bin" matches Spanish "composta", not real composters). - **`super_search=True`** — slow, high-quality. ~1–2s. Run when the user describes what they want rather than naming it: cross-language ("Schraubenzieher Set" → real screwdriver sets even without German catalog entries), attribute-style ("small metal part with a flat head"), or any case where the default returns junk. Embeds the query with voyage-3-large, takes the cosine top-50 over the corpus (with an exact-name precision boost for part numbers), then a cross-encoder reranks them. The two modes are mutually exclusive in practice — pick one based on whether the user knows the product's name or is describing it. Use this when the user asks to find a specific product or browse products matching a query. Prefer over `search_stores` when the intent is product-led ("find a drill") rather than store-led. Use `get_product` afterwards if the user wants full details for one specific result. Read-only. No authentication. Rate-limited to 100 requests/hour per IP. Args: query: Free-text search term. In default mode, treated as keywords (each word matched against product text). In `super_search=True`, treated as a natural-language description. min_price: Lower bound on price in EUR. Omit for no lower bound. Null-priced rows are NOT excluded by this filter — pass `has_price=True` if you need only priced listings. max_price: Upper bound on price in EUR. Omit for no upper bound. Tip — narrow by budget: `min_price=10, max_price=50, sort_by="price_asc", has_price=True`. Products without a listed price (a large fraction of the scraped catalog) sort last under either price ordering and are kept in results unless `has_price` filters them out. tags: Comma-separated tag filter (e.g. "electronics,bluetooth"). Tags are AND-ed together. store_id: Restrict results to a single store. Use the integer `id` from `search_stores` results. sort_by: One of `price_asc`, `price_desc`, `name_asc`, `newest`, `oldest`. Omit to use the default search-relevance ranking. has_price: When True, exclude products without a listed price (~most of the scraped catalog). Use this for competitive pricing or budget-bounded shopping. When False, return only null-priced listings (rarely useful). Omit to include both. semantic: Legacy flag. Pure vector ordering, ~250ms. Mostly superseded by `super_search=True` (which uses the same vector retrieval plus a cross-encoder rerank for materially better ordering at the cost of another ~700ms). Keep using it only if you specifically want vector retrieval *without* the rerank. super_search: **Enable for natural-language / "describe what I want" queries.** ~1–2s. Embeds the query with voyage-3-large, takes the cosine top-50 (with a precision boost for exact-name matches like part numbers / SKUs), then a cross-encoder reranks them. Use whenever the user is describing rather than naming — cross-language ("Schraubenzieher Set"), attribute-style ("small black metal bracket"), or any case where the default keyword path returns junk. Don't combine with cheap browse-style queries where the user typed an exact product name — keyword default is faster there. On `relevance_score` here: better than the bi-encoder cosine, but still not a "did I find what the user wanted" gauge. Behavior to expect: gibberish or fully-off-topic queries cap around 0.35; loosely-related catalogue clusters can score 0.7+ even when no item truly matches (a "ceramic vase" query in a catalog with no vases but many ceramic flowerpots will still score high). **Read the product names** before claiming a match. The score is most useful as a relative signal within one result set — a sharp drop between rank N and N+1 marks where the catalog stops being useful for this query. limit: Max results (1–100, default 20). Larger limits are slower and consume rate budget faster. offset: Skip this many results before returning. Use for pagination (offset += limit on each follow-up call). Returns: A list of products. Each includes `id`, `name`, `price`, `currency`, `url`, `description`, `store` (id/name/address), `tags`, `images`, a canonical `partle_url`, and `relevance_score` (cosine similarity 0–1 between the query and the product's embedding when a query was provided; `None` otherwise). **Always share `partle_url` with the user so they can view the listing.** Caveat on `relevance_score`: it is monotonic *within a single search result set* (useful for spotting a big drop-off between rank 3 and rank 4), but its absolute value is not well-calibrated across queries — most results land in 0.55–0.80 regardless of whether the catalog has truly relevant items. Don't infer "this is a great match" from a 0.75 score alone.
    ConnectorNo auth
  • Create a new product listing on Partle. Authenticated. Prefer **OAuth**: connect once via the consent flow on claude.ai (or any MCP client that supports OAuth) and the bearer token is attached automatically — no `api_key` parameter needed. **Fallback**: pass an `api_key` (prefix `pk_`, generate at /account) for programmatic or non-OAuth clients. Required OAuth scope: `products:write`. Use when the user wants to add an item for sale. For edits to an existing product, use `update_product` instead. **Images.** This tool creates text fields only — no image arg. Do **not** try to pass image bytes through a tool argument; phone-sized payloads blow past conversation context limits. The response includes a one-shot ``upload_url`` (signed, ~15 min TTL, bound to this product and your authenticated user). To attach an image from your code-execution sandbox, do **one** PUT request — no auth headers needed, the URL itself carries the credential: requests.put(result["upload_url"], data=open("/path/to/photo.jpg", "rb").read(), headers={"Content-Type": "image/jpeg"}) The bytes flow Python → HTTP body → Partle, never through the conversation. The URL works once and expires fast. Alternative if you don't have local bytes but have a public image URL: call ``upload_product_image(product_id, image_url=...)`` instead. **Duplicate prevention.** Same user, same product name (case- and whitespace-insensitive) returns 409 with `existing.id`, `existing.url`, **and a fresh `upload_url`** for that existing product — so if the user is just retrying with a photo, you can attach it directly to the existing listing without having to create or pick anything new. You can also call `update_product` to change fields. Don't retry blindly. **Idempotency.** Pass `idempotency_key` (any unique string per logical create — UUID or hash of the source listing) and a retry after a network failure returns the original response instead of creating a duplicate. Reusing a key with a different payload is a 422. Args: name: Product name. Required, 1–200 chars. description: Long-form product description. Optional. price: Price in whole currency units, **not** cents (e.g. ``15.99`` means €15.99). Max 100000. Omit for "ask the seller". currency: Currency symbol. Defaults to `€`. Use `$`, `£`, etc. url: Link to the merchant's product page. Optional but recommended. store_id: ID of the store this product belongs to. Omit for a personal listing not tied to any store. listing_type: ``in_stock`` (default) when the seller has the item and it can be bought now. ``tentative`` when they do not stock it and want to measure interest first — such a listing is kept out of normal search results and instead collects "I need this" presses. Only use ``tentative`` if the user explicitly said they are gauging demand; an item that is merely out of stock today is still ``in_stock``. If the user is looking to *buy* something nobody sells, use `create_buy_request` instead — that is the demand side and it is a different tool. idempotency_key: Optional retry-safety token. Unique per logical create. Send the same key on retries to get the same response. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: The created product record including its new `id` and canonical `partle_url`. Share `partle_url` with the user. Returns ``{"error": ...}`` on auth, dedup, or validation failure (dedup also returns ``{"existing": {"id", "name", "url"}}``).
    ConnectorNo auth
  • Estimate sourcing cost for a product based on fabric price, supplier pricing, and order quantity. USE WHEN: - User asks "how much would it cost to make 1000 t-shirts" - User needs a rough cost breakdown for budgeting - "ballpark cost to produce [quantity] [product] in China" - "budget estimate / sourcing cost / cost per piece for [product]" - "fabric cost + lead time estimate for [product]" - "how much to make [product] in [province]" - "rough quote / pricing range" - "can I make [product] for under $X per piece" - "多少钱 / 成本估算 / 报价 / 预算 / 做一批 [品类] 要多少钱" - "[省份] 做 [品类] 的成本大概多少" WORKFLOW: estimate_cost → optionally search_fabrics first to identify specific fabric_ids for accuracy → then recommend_suppliers for ready sources. RETURNS: { product, quantity, province, fabric_options: [{name, min_rmb, max_rmb, weight_gsm}], fabric_cost_per_meter, supplier_availability: { total_suppliers, avg_lead_time_days }, note } EXAMPLES: • User: "Rough cost to make 1000 cotton t-shirts in Guangdong" → estimate_cost({ product: "t-shirt", fabric_category: "knit", quantity: 1000, province: "Guangdong" }) • User: "What's the budget range for 5000 hoodies" → estimate_cost({ product: "hoodie", quantity: 5000 }) • User: "做 2000 件羽绒服大概多少钱" → estimate_cost({ product: "down jacket", quantity: 2000 }) ERRORS & SELF-CORRECTION: • fabric_options empty → no matching fabrics for the product term. Call search_fabrics directly with broader composition or widen the category, then re-estimate. • supplier_availability.total_suppliers = 0 → drop province filter or broaden product term. • Rate limit 429 → wait 60 seconds; do not retry immediately. AVOID: Do not present the output as a binding quote — always say "estimate based on database averages, not binding". Do not try to calculate per-piece cost from fabric alone — include labor, trim, margin externally. Do not use for detailed BOM costing — use search_fabrics + get_supplier_detail manually. CONSTRAINT: These are estimates based on database averages, NOT binding quotes. Always clarify this to the user. Fabric cost is per meter (typical usage: 1-3m per piece). NOTE: Cost accuracy improves when you provide a specific fabric_id via search_fabrics first. Source: MRC Data (meacheal.ai). 中文:按面料均价 + 供应商供货能力估算 [品类] 的生产成本区间。仅供参考,非正式报价。
    ConnectorNo auth
  • Search affiliate shopping products and get ranked recommendations. Only `query` is required — a natural-language description of what the shopper wants (e.g. "a warm waterproof jacket for winter hiking"). Add fields to narrow results: - keywords: exact product terms. Use a SPECIFIC product type ("women running shoes", "stainless steel knife set"), not a bare generic noun ("shoes", "knife") — a generic keyword can surface the wrong audience. Audience/occasion go in `query`. - commerce_l2s: category ids to restrict to (read the `commerce://categories` resource for valid values — that taxonomy is for shopping only). - max_price / min_price: price band (USD). platforms: ["amazon"|"walmart"] (empty=all). - intent: ranking preset — "cheapest" | "best_discount" | "top_rated" | "best_value". - require_commission: only return products that pay a commission (default: server setting, normally on). Pass false to include zero/unknown-commission products. - limit: max results (default 10). `agent_id` is REQUIRED: your registered, active agent id. It is the attribution key and the access key — a missing/blank or unregistered agent_id is rejected (no anonymous use). Returns {count, products:[{offer_id,title,brand,price,rating,...,buy_url,product_url}]}. Each product's `buy_url` is ALREADY the trackable affiliate buy link — hand it to the shopper directly. (`product_url` is the plain product page.) There is no separate resolve/click step.
    ConnectorNo auth
  • Buy a specific product version. THIS SPENDS MONEY from the agent wallet: the wallet must cover the price (insufficient_wallet_balance otherwise; check it with kenwea.wallet.getBalance), and the price counts against the operator's daily budget (budget_exceeded when it would go over). productVersionId is a product VERSION id, not a product id; find it with kenwea.marketplace.search or kenwea.marketplace.preview. A version that has not passed the sandbox is refused with sandbox_not_approved. A price of 1000 USDT or more waits for operator approval and moves no money until then. Returns a LicenseID; to put the product to use, call kenwea.marketplace.install with it.
    Connector
    Destructive
    No auth