Seller Assistant

Edit a team repricer strategy. First retrieve current strategies, translate the requested change into supported settings, and show the complete proposed behavior and safeguards. Require explicit confirmation. Pricing-relevant changes immediately enqueue repricing for affected active SKUs; setting this strategy as default replaces the previous default.

edit_repricer_strategy

Find purchase orders belonging to the current team. Filter by supplier ID or partial supplier name, exact po_id, and/or status. Filters are combined. With no filters, return the 10 newest team orders. Always show purchase_order_url as a clickable link when presenting an order. Use page and limit only when the user needs more filtered results; do not retrieve every page automatically.

find_purchase_orders

Confirm that Seller Assistant is connected and return the authenticated user, current team, active plan, connected Amazon marketplaces, and remaining feature quotas. This tool requires no arguments, has no side effects, and should be used when the user asks whether the connection works or which account is active.

get_account_context

Read the status and the results of a brand analysis started by start_brand_analysis. Poll this about every 20 seconds until is_terminal is true, and report progress only when the status changes. When it finishes, summarize the brand economics: average selling price, monthly revenue, seller count, how much Amazon itself sells, and the resale potential score. Set include_products to true to also list the brand's best products by sales rank. The suppliers in the response are ones the team ALREADY has for this brand. To find new suppliers, call search_supplier with the same brand name.

get_brand_analysis

Return the open next actions of the current team: the tasks that need the user, for all tools (purchase orders, repricer, setup) or for one tool. With target (for example po:123 or repricer:{team_id}), return the rows of that one object. Each row has a title, a reason, a count, a priority (lower comes first) and its actions. An action with an mcp hint tells you how to do it: follow its steps (each step names an MCP tool and its known arguments), ask the user for any value you do not know, and get confirmation before a write. When mcp.user_only is true, only the user can do it: explain it and give the route. A row resolves itself when its data changes, so do not mark it done. Hide or snooze a row with update_next_action only when the user asks. Hidden and snoozed rows are not returned. Use this tool when the user asks what to do next or what needs attention.

get_next_actions

Get products from a completed price list selected by list_id, exact description, or latest. Duplicate descriptions resolve to the newest matching list; with no selector, the newest team list is used. result_set defaults to potential_leads, which returns only items meeting Seller Assistant's default criteria (minimum profit and ROI, minimum estimated monthly sales except in the local environment, Amazon not currently in the Buy Box, and excluding AI-sourced matches). Use custom with filters to find better or different opportunities using the same advanced filters as the PLA dashboard. Custom requests must contain at least one filter or a search term; the complete dataset is not returned to the agent and remains available through price_list_url. Range filters use keys such as {"greater_equal":20,"less_equal":100}; text filters use {"checkboxes":["contains"],"inputs":{"contains":"value"}}; brand/category/size_type use {"checkboxes":["is"],"inputs":{"is":["value"]}}; and warning, restrictions, source, preference, return_rate, out_of_stock, alert, and tags use the dashboard's include/exclude values. Call check_price_list_status first to make sure the list has finished processing. Use the returned identifier/supplier_sku and cog to create purchase orders for some or all of these leads afterwards. When presenting results, always show the ASIN and make it a clickable Markdown link using amazon_url, for example: Product: Title — ASIN: [B012345678](amazon_url). Also show price_list_url as a link to the full analysis. If count is greater than zero, ask the user whether they want to create a purchase order from some or all of the returned leads. Number the displayed leads so the user can select them by number; also accept selection by ASIN, supplier SKU, or product name, with an optional quantity for each.

get_price_list_leads

Research one Amazon product by ASIN. This is the tool to use when a user asks about a product, whether it is worth selling, how well it sells, what it costs in fees, or who competes on it. It returns the title, brand, sales rank, estimated monthly sales, category, Buy Box and FBA/FBM prices, Amazon fees, offer and competitor counts, package size and weight, and warning flags such as meltable, fragile, oversize, hazardous or adult. Add cost_of_goods to also get the profit, ROI and margin at the current Buy Box price. Leave include_price_history true to also get the 30, 90, 180 and 365-day averages for price, sales rank, offer count and reviews, which show whether the product is stable or falling. For the team's own selling permission on the ASIN, use check_product_restriction.

get_product

Get a bounded page of team repricing history and explain price decisions. History covers the last 32 days only. Filter by SKU, ASIN, marketplace, status, mode, error, period (24h, 7d, 30d), or date. Without a date or period, the window is the last 32 days. The search term matches the start of an ASIN or a SKU. Defaults to the 20 newest events; maximum 100. Do not retrieve all pages automatically.

get_repricer_history

Return compact team repricer performance and health: coverage, modes, Buy Box win rate, the open repricer next actions (the issues to fix, with product counts), operation success, and price movements. Defaults to this week; supports day, week, month, or an explicit date range. History covers the last 32 days only, so an earlier start date counts from the 32-day limit.

get_repricer_performance

Get paginated per-ASIN results for a completed restriction check. Each result is ALLOWED_TO_SELL, APPROVAL_REQUIRED, NOT_ELIGIBLE, or ERROR. Call check_restriction_status first. Use page and limit for large batches; never request or present more pages unless the user needs them.

get_restriction_check_results

Get the suppliers found by a completed search selected by search_id, exact brand, or latest. Duplicate brands resolve to the newest matching search; with no selector, the newest team search is used. Call check_supplier_search_status first to make sure the search has finished. Each result carries is_added, which says whether the team already saved that supplier. For a result where is_added is false, offer to save it with add_supplier so the user can order from it.

get_supplier_search_results

List the brands the team has already analyzed, newest first. Use this whenever the user refers to an earlier analysis rather than asking for a new one — "the brand I checked last week", "my brand analyses", or a brand name they have looked at before. Each row carries the brand_analyzer_id that get_brand_analysis takes, so a finished analysis can be read again for free. Search for the brand name here BEFORE calling start_brand_analysis, because starting a new one spends a brand quota unit. A row whose status is not Download or Failed is still running; poll get_brand_analysis with its ID. Returns at most 50 rows.

list_brand_analyses

List the price lists the team has already analyzed, newest first. Use this whenever the user refers to an earlier analysis rather than a new one — "the list I uploaded last week", "my Webstaurantstore list", or simply "my price lists". Each row carries the list_id that get_price_list_leads takes, so an old analysis can be read again without re-uploading the file and spending another quota unit. leads_count is how many profitable products the analysis found. A list whose status is not Download or Failed is still processing; poll check_price_list_status with its list_id. Returns at most 50 rows.

list_price_lists

List the current team's Amazon SKUs with their repricer state. Call this before any repricer product write, because assign_repricer_strategy, set_repricer_product_status and set_repricer_product_limits all take the repricer_sku_id values this tool returns. Use issue_types to find products that need attention (the same issues as the repricer next actions): no_cost (no landed cost, so cost-based price floors cannot be computed), no_postage (an FBM SKU with no FBM postage, so it has no profit and no cost-based price floor; set fbm_postage_cost with set_sku_costs), on_floor (the last run stopped at the min price), capped (the last run stopped at the max price), failed (Amazon did not accept the last price), price_war (repeated competitor undercutting) and inactive (an inactive listing with stock). listing_state gives one listing state: active (an active listing with stock), out_of_stock or inactive. A next action row gives the issue_types and listing_state that read its products. fulfillment_channel is FBA or FBM and selects the profit formula. price, min_price and max_price are item prices; for FBM, my_shipping is the shipping charge of our offer and the buyer pays price + my_shipping. FBM profit = price + my_shipping - referral fee - closing fee - product cost - prep - fbm_postage. buybox_status is out_of_stock when the SKU has no stock (an out-of-stock offer cannot hold the Buy Box), else the last known status; unknown means no Buy Box data yet. market_data_updated_at gives the age of the data. The buybox_statuses value stale finds SKUs whose data is older than the stale limit (2 hours by default); the repricer refreshes data at least every 30 minutes, so old data is a problem on the Seller Assistant side, not a task for the user. A row with repricer_sku_id set to null is not enrolled in the repricer and cannot be written to. Returns at most 100 rows; use repricer_url for the full list.

list_repricer_products

List the current team's repricer strategies and the supported strategy-building capabilities. Call this before recommending or creating a strategy. Use the user's business intent, risk tolerance, cost data, inventory, competition, and profit goals to choose settings; the user does not need to know internal action or condition names. Returns at most 50 strategies and never retrieves repricer products.

list_repricer_strategies

List the bulk restriction checks the team has already run, newest first. Use this when the user refers to an earlier check rather than asking for a new one — "the restrictions I checked yesterday", or "my restriction checks". Each row carries the restriction_id that check_restriction_status and get_restriction_check_results take, so a finished check can be read again without re-running it and spending restriction quota. The counts summarize the outcome: how many ASINs the team may sell, how many need approval, how many are not eligible, and how many failed. A row where is_terminal is false is still running. Returns at most 50 rows.

list_restriction_checks

List the current team's SKUs with their costs, and where each cost came from. Call this before set_sku_costs, both to get the sku_id values and to show the user what changes. cost_source explains each cost: override is a cost the user typed in and it wins over everything else, po_linked comes from a linked purchase order, po_wac is a weighted average across purchase orders, and none means the SKU has no cost at all. Use cost_source=none to find the SKUs a user must still cost. Those are the same SKUs the repricer reports with the no_cost problem, and they cannot use a break_even, target_roi or target_margin price floor. Returns at most 100 rows.

list_sku_costs

List the brand searches the team has already run, newest first. Use this whenever the user refers to an earlier search rather than a new one — "the suppliers I found for Acme last week", or simply "my supplier searches". Each row carries the search_id that get_supplier_search_results takes, so an old search can be read again without spending another quota unit. A search whose status is not Done or Failed is still processing; poll check_supplier_search_status with its search_id. Returns at most 50 rows.

list_supplier_searches

List the suppliers the current team has saved. Call this to find the supplier_id that create_purchase_order needs, and call it before add_supplier to check whether the supplier already exists. Use search to match a name or a website. Returns at most 50 suppliers.

list_suppliers

Manually match one or more Amazon ASINs to global product identifiers (12-digit UPC or 13-digit EAN) for the authenticated team. This uses Seller Assistant's existing product-matching rules: the ASIN identifies or creates the internal product for the selected marketplace, the UPC/EAN is linked as a manual global identifier, and an exact blacklist entry for the pair is removed. Before calling, show every ASIN → UPC/EAN pair and marketplace to the user and ask for confirmation.

match_product_identifiers

Set the Minimum Advertised Price on one or more ASINs. MAP is the lowest price a brand allows a seller to advertise. It belongs to the ASIN, so it applies to every SKU the team lists for that ASIN — it cannot be set for one SKU alone. Take the repricer_product_ids from list_repricer_products; they are the ASIN-level IDs, NOT the repricer_sku_ids. MAP is an item price: for an FBM SKU with a shipping charge, the buyer pays the item price plus shipping. A strategy respects MAP unless its Ignore MAP setting is on. Setting MAP on an ASIN with no strategy stores the number but nothing enforces it yet. Send map_price as null to clear it. To set the hard price floor and ceiling of a SKU, which are different numbers with a different owner, use set_repricer_product_limits. Show the ASINs and the new MAP to the user and get explicit confirmation before calling.

set_repricer_map_price

Set the hard price floor and ceiling of one or more SKUs. min_price is the lowest price the repricer may ever set and max_price is the highest; it never crosses either. Take the repricer_sku_ids from list_repricer_products. These are per-SKU numbers, so each SKU of an ASIN can carry its own. Only the fields you send change; the other keeps its value. Send null to clear one. Two neighbouring settings live elsewhere. How a floor is CALCULATED (break-even, target ROI, target margin) is a strategy setting — use create_repricer_strategy or edit_repricer_strategy. The Minimum Advertised Price belongs to the whole ASIN, not a SKU — use set_repricer_map_price. Show the exact new values to the user and get explicit confirmation before calling. A limit change re-prices the affected active SKUs within seconds.

set_repricer_product_limits

Turn repricing on or off for products. This is a simple switch with two positions, the same switch the Seller Assistant app shows. mode=live turns repricing on and the repricer sends new prices to Amazon. mode=off turns it off and the product keeps its current price. Get the repricer_sku_ids from list_repricer_products. Before turning products on, confirm they have a minimum price, because a product with no price floor can be repriced down to whatever the strategy returns. The plan caps how many products can be on at once, so read limit_reached in the response: above zero means some products stayed off. Always ask for explicit confirmation before calling.

set_repricer_product_status

Set the cost a team paid for its SKUs. The cost is stored as a manual override and wins over any cost derived from a purchase order, so use this when the user bought outside a tracked purchase order or wants to correct a cost. Name the target in one of two ways. sku_ids changes exactly those SKUs; get them from list_sku_costs. asin changes every SKU the team lists for that ASIN, which is often several, so before calling with an asin you MUST run list_sku_costs first, show the user every SKU the cost will apply to, and get confirmation. Add marketplace_id to keep an ASIN change inside one marketplace. Send null for a cost field to remove the override and let the purchase-order cost return. Changing a cost re-prices any repricer product that uses a cost-based price floor, and it corrects the historical profit already reported for the SKU.

set_sku_costs

Analyze a whole Amazon brand: how many products it has, what they sell for, how much revenue they make, how many sellers compete, how much Amazon itself sells, and whether the brand is worth reselling. Use this when a user asks about a brand rather than one product. The analysis runs in the background, so poll get_brand_analysis about every 20 seconds with the returned brand_analyzer_id until the status is Download or Failed. An analysis costs one brand quota unit, but a brand the team analyzed in the last 30 days is returned instead of a new one and costs nothing. Call list_brand_analyses first when the user may have analyzed this brand before.

start_brand_analysis

Start analyzing a supplier price list to find profitable Amazon resale opportunities. When mapping a large source file, inspect only a few representative rows to confirm the required fields and mapping; do not validate or analyze every row yourself. Provide the mapped items — each needs at least an "identifier" (UPC, EAN, or ASIN) and "cog" (cost of goods); "title", "supplier_sku", "price_type" and "supplier_case_quantity" are optional. "cog" is a per-unit price unless "price_type" is "case", in which case it is the price of a whole case and "supplier_case_quantity" is required. Optional expenses can override the user's saved PLA logistics, prep, tax, VAT, Amazon-fee, and miscellaneous-cost settings for this analysis only. For VAT-based non-US marketplaces, use vat_scheme, the applicable VAT rate, cog_incl_vat, and vat_on_seller_scheme/vat_on_seller_fees. Analysis runs in the background and can take several minutes. After this tool succeeds, automatically poll check_price_list_status about every 20 seconds, show progress when it changes, and call get_price_list_leads when processing finishes.

start_price_list_analysis

Start an asynchronous Amazon selling-restriction check. An active Amazon seller account connection for the marketplace region is required. Provide exactly one source: either a batch of ASINs plus marketplace_id, or price_list_item_ids previously returned by get_price_list_leads. For PLA items, the marketplace and ASINs are resolved automatically and all items must be from the same marketplace. After starting, poll check_restriction_status until terminal, then call get_restriction_check_results. If the user is referring to a check they already ran, use list_restriction_checks instead of starting a new one.

start_restriction_check

Search the open market for wholesale suppliers and authorized distributors for a brand. This does not search the team's saved suppliers; use list_suppliers for those. The search runs in the background — poll check_supplier_search_status about every 5 seconds with the returned search_id, then call get_supplier_search_results once it finishes. A search consumes one supplier_finder quota unit, but a repeat search for the same brand and context in the last 24 hours is returned instead of a new one and costs nothing. Call list_supplier_searches first when the user may have searched this brand before. Add the optional context fields when the user gives them, because they make the results more relevant.

search_supplier

Snooze, hide or show again one next action row, only when the user asks for it. snooze hides a snoozable row for 7 days. hide hides a dismissible row until its data changes (a new item brings it back). restore shows a hidden or snoozed row again. The change applies to the whole team, the same as the row menu in the app. Take id (for snooze and hide) or key (for restore) from get_next_actions. Do not use this tool to mark a fixed row: a row resolves itself when its data changes.

update_next_action