Analyze an Amazon Merch or KDP niche with the Flying Research backend. Returns total match counts, bounded design and best-seller samples, and MBA, Redbubble, and Spreadshirt keywords. Ranks are dated stored observations; null currentBsr means no verified rank within 24 hours. Summary omits historical time series by default. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read synonyms for a keyword and language. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate keyword tags from the stored keyword service. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search the keyword service using the app searchType. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read Amazon clothing product ranks for selected ASINs. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search historical Amazon archive snapshots by archive_date and filters. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read the research trend overview. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search research brands with an optional minimum brand occurrence. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read known official brands for a marketplace and product. limit bounds products across every brand group; offset continues within the same backend page. Follow nextOffset automatically when needed within the remaining record allowance. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read a stored Amazon product detail by ASIN. Check the returned marketplace/product before attributing data. PRODUCT_NOT_FOUND means this connected catalog has no stored detail, not that the product is absent on Amazon. For Find Copycats, use the resolved marketplace/product to run an exact-ASIN Live Research lookup, wait for completion, then retry this stored lookup to verify ingestion. Do not treat permission or server failures as a missing product. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read a bounded sample of ranked stored listings matching every keyword phrase in title or brand. Specify marketplace to restrict the market; omission searches all markets. Results are stored observations, not total live Amazon competition. Use niche analysis for aggregate niche metrics. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read designs related to an ASIN by brand. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read designs related to an ASIN by keywords. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read saved favorites and their IDs for subsequent update/removal. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Replace the tag on a saved favorite using its favoriteId. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Remove a saved favorite by favoriteId. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read the current account favorite tags. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Find niche keyword ideas for a marketplace. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read upcoming calendar events and related designs. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read country-wide Google trends and keyword-specific related topics. Partial results identify unavailable provider sections; missing sections do not mean no interest. Country-wide trending searches are not keyword search-volume data. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search stored deleted-design observations. Deletion is not evidence of an infringement decision. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Send selected image links to the specified email address. Sends an email; use only when the user requests this. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search stored Best Sellers or Seller Trends. Defaults to specific phrase search for targeted niches; explicitly choose another search_type for broad exploration. Results include matched fields and dated BSR evidence, not guaranteed current ranks. Filters apply to this request; use page to load more. statistics summarizes only the returned page. For selection, pass returned product identities to subsequent tools. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Compare one to three selected products, with stored BSR history and sample statistics. Validates ASIN, marketplace and product identity before attributing data. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Save one or up to 50 Amazon, Redbubble or Spreadshirt products to Flying favorites. Use identities returned by research tools, with an optional tag per product (defaults to no_tag). Reports each attempted product and partial progress; stops on the first failure. Does not remove favorites. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read, add or remove product identities on your persistent MCP research board. Separate from the browser page ASIN Board. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Load the relevant Flying workflow before research. Use seller-watch to compare an ASIN with its niche using live research, competitor histories and observed 12-week rankings. Use find-copycats to search Best Sellers plus live featured and newest results and visually review similar artwork. Follow every required stage, the connected server limits and the completion check in the returned skill; report partial or blocked stages explicitly. This returns instructions only, not completed research or a scheduled task. The similarly named brand-snapshot and text-candidate helpers do not execute these full workflows.

Return actual viewable product images as native MCP image content, labelled with each ASIN. Use for the original and candidate visual inspection in Find Copycats; image URL strings alone are not visual evidence. Supply exact product identities, optionally imageUrl from a returned live row when no stored image exists. Stored images take precedence; caller URL associations are explicitly unverified. Bounded batches and image sizes follow the returned policy. Continue further candidate batches within the remaining record allowance, reusing already-delivered images. Returns per-image delivery/failure/not-attempted status; delivery does not mean reviewed and no similarity score is calculated. Public Amazon CDN images only; no writes or paid AI generation. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Calculate Seller Watch historical comparisons across the bounded stored niche cohort in ONE call. Prefer this over handpicking peers or fetching histories individually. Requires target identity, primary niche phrase, IANA timezone and remaining recordBudget including the target. Bulk-loads dated raw histories internally and returns server-calculated 12 completed weekly ranks, a separate current incomplete-week rank, matched recent-period median changes, all assessed peer titles/category exclusions in cohortReview, coverage and stopping reasons. Only known matching current category labels enter primary comparative statistics; unknown or incompatible labels remain explicit exclusions. Target-only ranks do not establish niche leadership. Uses stable ASIN ordering if limited; display examples do not limit the analyzed cohort. Stored full-text matches require relevance review; historical category compatibility is provisional. Does not run live research, niche analysis, save the board or schedule tasks; those skill stages remain separate. No raw history export or automatic allowance bypass. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Legacy brand-catalog snapshot helper. Use only for an explicitly requested stored brand-sample comparison, not ASIN-versus-niche Seller Watch. For niche performance, first load flying_research_skill with skill seller-watch and follow that workflow instead. This helper resolves a Merch seed, reads a bounded exact-brand page in the same marketplace/product, compares it with the previous account snapshot and saves a replacement snapshot for that scope. Reports first-seen/not-seen-in-sample and price/stored-BSR changes; does not run live niche research or reconstruct weekly niche ranks. Brand grouping does not verify seller ownership; KDP author catalogs are unavailable. Keep page and limit unchanged on repeat comparisons. No recurring task is scheduled. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Optional stored text-candidate helper with actual image previews, not the complete Find Copycats workflow. First load flying_research_skill with skill find-copycats for Best Sellers, live featured, live newest and image-review stages. Resolves the original, searches the same marketplace/product and returns metadata/text evidence. includeImages defaults true: native MCP image content for the original plus a bounded first candidate subset, labelled by ASIN, reusing loaded records. Use flying_research_images for remaining candidates; the preview never defines full review coverage. Set includeImages false for metadata only. searchTerm can supply a distinctive phrase. Does not run the required live searches, perform visual assessment or measure similarity. No takedowns or messages are sent. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate a private temporary CSV download for up to 50 selected products. Reads verified records; protects against spreadsheet formula injection. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Start independent real-time research using the existing authenticated live research server. Poll with flying_live_status. Fetches the first requested page, up to 48 products. After it finishes, automatically use flying_live_next with the latest jobId and the same filters when more results are needed and continuation.available is true. One page job per account; at most 3 pages / 144 unique products per search. Actual counts depend on available listings, duplicates and scrape success. Real-time research limit: 3 consecutive pages and 144 unique products per search, fetched one page at a time (up to 48 products per page). After the current page finishes, automatically call flying_live_next with the latest jobId when continuation.available is true and more results are needed. No new user next message is required within these limits. Respect any tighter workflow record budget. Do not start parallel jobs, change filters, or restart the search to bypass the page or product limits. Status waits briefly for completion by default. If it is still running, wait at least retryAfterSeconds before polling the same jobId again. An HTTP 429 does not stop the worker: wait for Retry-After/resetAt, then resume the same job; do not restart or skip the search. A terminal=true response ends the current page, even when partial=true. Poll or stop a running page job; do not request its next page before it finishes. Report actual page and cumulative counts; duplicates and scrape failures can reduce them.

Fetch ONE next page of the same search. Automatically call after the current page finishes when continuation.available is true and more results are needed within the search limits and remaining workflow budget. Use the latest jobId. Never change filters or start a replacement search to bypass the three-page / 144-product total. Poll this page to completion before requesting another page. Real-time research limit: 3 consecutive pages and 144 unique products per search, fetched one page at a time (up to 48 products per page). After the current page finishes, automatically call flying_live_next with the latest jobId when continuation.available is true and more results are needed. No new user next message is required within these limits. Respect any tighter workflow record budget. Do not start parallel jobs, change filters, or restart the search to bypass the page or product limits. Status waits briefly for completion by default. If it is still running, wait at least retryAfterSeconds before polling the same jobId again. An HTTP 429 does not stop the worker: wait for Retry-After/resetAt, then resume the same job; do not restart or skip the search. A terminal=true response ends the current page, even when partial=true. Poll or stop a running page job; do not request its next page before it finishes. Report actual page and cumulative counts; duplicates and scrape failures can reduce them.

Read a live research job owned by this account and OAuth client. Returns a readable product table and the original structured data. Present the summary and table to the user, with clickable product titles and formatted numbers. Analyze the returned products and state the number actually analyzed. Use status and limitations to explain any collection limit, scrape failure, cancellation, or timeout. A completed job covers the current page; follow continuation.message and automatically continue when available and needed within the search limits and remaining workflow budget. A larger Amazon search total does not mean analysis of the returned products failed. Do not claim exhaustive Amazon coverage or describe a collection cap as an account request-rate limit. Real-time research limit: 3 consecutive pages and 144 unique products per search, fetched one page at a time (up to 48 products per page). After the current page finishes, automatically call flying_live_next with the latest jobId when continuation.available is true and more results are needed. No new user next message is required within these limits. Respect any tighter workflow record budget. Do not start parallel jobs, change filters, or restart the search to bypass the page or product limits. Status waits briefly for completion by default. If it is still running, wait at least retryAfterSeconds before polling the same jobId again. An HTTP 429 does not stop the worker: wait for Retry-After/resetAt, then resume the same job; do not restart or skip the search. A terminal=true response ends the current page, even when partial=true. Poll or stop a running page job; do not request its next page before it finishes. Report actual page and cumulative counts; duplicates and scrape failures can reduce them.

Request cancellation of your live research job. The worker stops its upstream socket on the next heartbeat; poll status to confirm. Real-time research limit: 3 consecutive pages and 144 unique products per search, fetched one page at a time (up to 48 products per page). After the current page finishes, automatically call flying_live_next with the latest jobId when continuation.available is true and more results are needed. No new user next message is required within these limits. Respect any tighter workflow record budget. Do not start parallel jobs, change filters, or restart the search to bypass the page or product limits. Status waits briefly for completion by default. If it is still running, wait at least retryAfterSeconds before polling the same jobId again. An HTTP 429 does not stop the worker: wait for Retry-After/resetAt, then resume the same job; do not restart or skip the search. A terminal=true response ends the current page, even when partial=true. Poll or stop a running page job; do not request its next page before it finishes. Report actual page and cumulative counts; duplicates and scrape failures can reduce them.

Process comma-separated keywords with the same filtering, formatting, sorting and grouping rules as KW Workbench. No provider calls or saved-list changes. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Get KW Suggestions from Amazon, Google Search or Google Shopping. Up to 30 prefix/postfix combinations; requires Keyword Basic or Pro. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate live Spreadshirt tags for a keyword and marketplace. Results include provider counts when available. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read Google interest over time for up to five years. Requires Keyword Basic or Pro; values are relative interest, not sales. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read a bounded page of your saved KW Lists. Follow hasMore and nextOffset. Do not save a truncated list. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Create or replace complete owned keyword lists. Read first and preserve all categories and notes. Supply a new unique id when creating a list. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Permanently delete selected keyword lists belonging to your account. Only use when the user requests deletion. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Run Product Explorer for a keyword returned by your completed KW Search job. Returns observed ASIN ranks; requires Keyword Pro. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Continue KW Hunter from a completed seed KW Search job. Combine Amazon/Google suggestions with Spreadshirt tags and start enrichment of up to 25 related keywords. Requires Keyword Pro; poll the returned job using keyword search status. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Start KW Search using the authenticated live keyword service. Accepts up to 25 keywords; Free accounts may search one seed using cached data only with an increasing daily cooldown and no hard daily search cap. Google search volume, CPC and search-volume competition require Keyword Basic or Pro. Poll flying_keyword_search_status for permitted metrics and suggestions. Results are bounded and may be partial. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read your keyword job and its partial results. Google searchVolume.freshness states the latest observed month and whether metrics are historical; disclose historical dates and do not present them as current demand. Paid accounts can retry with refresh=true. Jobs belong to the account and MCP client that started them; results expire after 24 hours. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Cancel your keyword job. Poll status until stopped; the worker disconnects the upstream socket on its next heartbeat. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Search trademark records by exact word, contains, phrase check or applicant. Results are research evidence, not legal clearance. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read paginated trademark records filtered by office and application/registration dates. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read paginated trademark records filtered by office and application/registration dates. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read a trademark by its record ID or backend-supported identifier. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Check up to 50 search phrases with explicit office, class, status and mark-type filters. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Check one text for stored trademark matches. A zero count does not establish legal clearance. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Check listing text fields for stored trademark matches. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read the account trademark watchlist with current match counts. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Add words and territories to the account trademark watchlist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Replace the word and territory of an owned watchlist entry. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Remove selected entries from the account watchlist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read the account whitelist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Add words, notes and languages to the account whitelist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Update an owned whitelist entry. The legacy backend does not clear fields when given empty strings. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Remove selected entries from the account whitelist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read the account blacklist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Add words, notes and languages to the account blacklist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Update an owned blacklist entry. The legacy backend does not clear fields when given empty strings. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Remove selected entries from the account blacklist. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Start here for listing-generation requests when the method is unspecified. Read the method choices, required inputs, supported models and generation workflow. The default generation tool is flying_ai_listing_batch, including for one design. This is server configuration, not private settings or credit balance. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval.

Preview the total API token cost for designCount selected designs using the selected mode/model and saved platforms/languages. Returns current token balance and whether it is sufficient. Does not consume tokens. Generate using flying_ai_listing_batch by default, including for one design. Ask whether to proceed with this total and wait for explicit user approval before generation; settings and balance can change. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Read rows from a user-provided Flying standard .xlsx workbook without importing designs or consuming AI tokens. Supply workbook bytes, not a server/local path. Image Path cells are labels only: the client must access user-provided matching images for Vision. Select rows, choose the method and obtain approval of the total cost before generating. At most 100 rows per user message. Automatically follow nextRow within the remaining row allowance when needed; after reaching it, wait for a new next message for another batch. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Create a new downloadable Flying standard .xlsx workbook from generated draft listings and original image paths. Does not modify the source workbook, import designs, publish listings or consume AI tokens. At most 100 selected rows. Requires file-export permission. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Find designs in your owned upload library for AI listing generation. Use returned flyIds as entries in flying_ai_listing_batch, the default generation tool. Supports uploaded_from/uploaded_to for designs added to the upload library today, in the last 7 days or last 30 days. Translate relative dates using the user’s timezone: today starts at local midnight; last 7/30 days include today plus the previous 6/29 calendar days, ending now. If timezone is unknown, ask before resolving relative dates. Show the resolved date range and actual selected design count. Fetch at most 500 records per user message; automatically paginate within the remaining allowance when needed. Then obtain approval of the total token cost before generating. Does not generate or modify listings. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Default listing-generation tool for Vision, Keywords, and Template. Use for both one design (one entry) and multiple designs. Generate listing drafts for all selected designs in one batch. Batch billing rounds once across all entries. Always preview first without expectedTokens, display the total, and wait for explicit approval. Then submit the same entries, mode and model with userConfirmed true and expectedTokens equal to the approved requiredTokens. Never split an approved batch into individual calls. No automatic retries; interrupted work may have consumed credits. Generates drafts only. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate AI listing text from a library design image or a user-provided local image (file.name and file.image). topic and instructions are optional. Arbitrary image URLs are not accepted.Single-design compatibility tool. Use flying_ai_listing_batch by default for all new generation requests, including one design. Never use repeated calls to this tool for a multi-design selection. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval. On the first call omit expectedTokens: the tool returns a preview and does not generate or charge. Ask whether to proceed with the displayed deduction, STOP and wait for explicit user approval. Only then call again with userConfirmed true and expectedTokens set to the returned requiredTokens. Choosing a mode or asking to generate is not approval of an undisclosed price. For multiple selected designs, call flying_ai_listing_estimate with designCount equal to the selected count, show the total and wait for approval of that total; generate all selected designs in one flying_ai_listing_batch call using the approved total as expectedTokens. Do not split the batch into individual paid calls; their minimum charges differ. Changes to the selection, mode, model or price require a new preview and approval. Never guess expectedTokens or skip showing the message. Generates one draft for an owned upload-library flyId or a user-provided file/Excel row. Set file.name for direct file generation; Vision also requires file.image bytes. Never import local sources into Design Management. The server cannot read device paths or fetch image URLs. For files flyId is a caller reference, not a library record. Uses your saved AI Listing settings (platforms, languages and text rules). Consumes Flying API tokens under the same billing/refund rules as the UI. Does not save, replace or publish listings. Partial/interrupted responses may already have consumed credits; do not automatically retry. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate AI listing text using mainKeywords and optional fillKeywords.Single-design compatibility tool. Use flying_ai_listing_batch by default for all new generation requests, including one design. Never use repeated calls to this tool for a multi-design selection. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval. On the first call omit expectedTokens: the tool returns a preview and does not generate or charge. Ask whether to proceed with the displayed deduction, STOP and wait for explicit user approval. Only then call again with userConfirmed true and expectedTokens set to the returned requiredTokens. Choosing a mode or asking to generate is not approval of an undisclosed price. For multiple selected designs, call flying_ai_listing_estimate with designCount equal to the selected count, show the total and wait for approval of that total; generate all selected designs in one flying_ai_listing_batch call using the approved total as expectedTokens. Do not split the batch into individual paid calls; their minimum charges differ. Changes to the selection, mode, model or price require a new preview and approval. Never guess expectedTokens or skip showing the message. Generates one draft for an owned upload-library flyId or a user-provided file/Excel row. Set file.name for direct file generation; Vision also requires file.image bytes. Never import local sources into Design Management. The server cannot read device paths or fetch image URLs. For files flyId is a caller reference, not a library record. Uses your saved AI Listing settings (platforms, languages and text rules). Consumes Flying API tokens under the same billing/refund rules as the UI. Does not save, replace or publish listings. Partial/interrupted responses may already have consumed credits; do not automatically retry. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.

Generate AI listing text using a template with keywords, blacklist and platform rules.Single-design compatibility tool. Use flying_ai_listing_batch by default for all new generation requests, including one design. Never use repeated calls to this tool for a multi-design selection. Before estimating or generating, check whether the user has chosen a generation method for this request. If unspecified, ask: "How would you like to generate your listing: Vision (from the design image), Keywords (from your keywords), or Template (from a keyword/rule template)?" Wait for their answer. Do not default to Vision or choose a method for them. If they already clearly requested a method, use it without asking again. Collect the inputs for their chosen method; do not invent keywords or a template. Handle multiple designs together automatically. Do not ask the user whether to use batching, mention batching or tool names, or show the pricing formula. Show only the selected design count, model and total API-token charge, then ask for approval. On the first call omit expectedTokens: the tool returns a preview and does not generate or charge. Ask whether to proceed with the displayed deduction, STOP and wait for explicit user approval. Only then call again with userConfirmed true and expectedTokens set to the returned requiredTokens. Choosing a mode or asking to generate is not approval of an undisclosed price. For multiple selected designs, call flying_ai_listing_estimate with designCount equal to the selected count, show the total and wait for approval of that total; generate all selected designs in one flying_ai_listing_batch call using the approved total as expectedTokens. Do not split the batch into individual paid calls; their minimum charges differ. Changes to the selection, mode, model or price require a new preview and approval. Never guess expectedTokens or skip showing the message. Generates one draft for an owned upload-library flyId or a user-provided file/Excel row. Set file.name for direct file generation; Vision also requires file.image bytes. Never import local sources into Design Management. The server cannot read device paths or fetch image URLs. For files flyId is a caller reference, not a library record. Uses your saved AI Listing settings (platforms, languages and text rules). Consumes Flying API tokens under the same billing/refund rules as the UI. Does not save, replace or publish listings. Partial/interrupted responses may already have consumed credits; do not automatically retry. Listing limit: 500 records total per user message. Automatically paginate sequentially within the remaining allowance when needed. Stop at the allowance and wait for a new "next" message for another batch. No parallel batches, alternate-tool workarounds, or whole-library files that bypass the allowance.