MCP App Store
by KeenEthics
Secnd icon

Secnd MCP Server for Claude and ChatGPT

By Secnd

Overview

Secnd is the inventory and cross-listing system resellers use to run one catalog across every marketplace they sell on. This connector gives Claude access to that catalog. It can search your products and the Secnd help centre, answer what is listed where, draft new listings from a description or a spreadsheet, suggest prices, publish to the marketplaces you have connected, update prices, delist, and mark items sold. It also reports sell-through and slow movers, so you can ask what is not moving and act on the answer in the same conversation. Everything is scoped to your own Secnd account through your own sign-in. You decide how far an agent may go: at secnd.ca/agents you grant a permission level — read, draft or edit — and a daily limit, and the server refuses anything beyond it. Tools that publish, reprice, delist or mark sold are marked destructive and confirm before acting; the rest are read-only. No tool can reach another seller's data. Desktop clients with file access can upload a folder of photos straight into drafts, so a bulk listing run that would take an afternoon in a browser fits in one conversation.

How to connect

  1. Claude: Open the Secnd listing in the Claude connectors directory and choose Connect. Claude listing Secnd exposes 28 tools here.
  2. ChatGPT: Open the Secnd listing in the ChatGPT app directory and choose Connect. ChatGPT listing Secnd exposes 28 tools here.

Tools

attach_uploaded_photos

ChatGPT
Attach photos uploaded through request_photo_uploads to the product, in the order given (first = cover photo). Each uploadKey is checked: it must belong to this product, the file must have reached storage, and it must be a readable image. Returns per key the image id and URL, or a failure code (UPLOAD_NOT_FOUND: upload it again with a new slot; CROSS_CATEGORY_KEY: wrong product). Safe to call again with the same keys: an already attached photo returns its existing image. Touches no marketplace.

bulk_update_listings

ChatGPT
Bulk-edit a set of the seller's marketplace listings. This is the general edit primitive — pair it with get_marketplace_listings (to pick the items) and always PREVIEW first. Select products by skus / titles / productIds. Set dryRun=true (default) to preview exactly what would change WITHOUT writing anything; set dryRun=false to apply. Applying re-pushes each listing through the normal update pipeline as a background job (returns a jobId) — no data is changed synchronously. Operation "reapply_section_mapping" (Etsy): re-derive each listing's Etsy shop section from the seller's section-mapping RULES and push it. Use for "a lot of my Etsy listings are in the wrong section — map them correctly per my section rules". In preview, each item shows targetSectionId (what the rules resolve to), any explicitSectionOverride, and whether that override is blocking the rule. Note: an explicit per-product section override wins over the rules on update; set clearSectionOverride=true to remove those overrides so the rules take effect. ALWAYS run dryRun first and confirm the plan with the user before applying.

create_listing_draft

ChatGPT
Create a DRAFT product in the seller's SECND catalog — the first step of listing an item. Fill in as much as you can from what the user shows or tells you (a photo in chat, a description): title, brand, size, condition, price, description. Pass public image URLs in imageUrls and SECND downloads them onto the product. Nothing is published anywhere by this tool — the draft sits in SECND until submit_listing pushes it to marketplaces. Returns the productId (for submit_listing) and a secndUrl the seller can open to review/edit the draft and add photos. Photos that are local files (you can read the file and make HTTP requests): create the draft without imageUrls, then request_photo_uploads + attach_uploaded_photos. Photos with no public URL and no file access: create the draft without images and tell the seller to add photos on the product page before submitting. For many items, wrap the calls in run_batch.

delist_listings

ChatGPT
End the live marketplace listings of products (the products stay in SECND). Without marketplaces, every live listing of each product comes down; with marketplaces, only those. DRY RUN BY DEFAULT: the first call returns, per product, which live listings would end; show that to the seller, then call again with dryRun false. Needs the seller to allow agents to remove listings (they turn it on in Secnd's agent settings); if not allowed the tool says so. Results: get_delist_results with the jobId.

fetch

ChatGPT
Fetch one record by an id returned from search: product:<uuid> gives the full product (fields, price, status, marketplaces, description, photo URLs); help:<slug> gives the help doc text. Read-only.

find_duplicate_products

ChatGPT
Find groups of likely duplicate listings in the seller's catalog (same product listed multiple times across or within marketplaces). Returns existing duplicate groups. Pass runDetection=true to also queue a background listing match over the whole catalog (only when the user asks to re-check): it merges the same item across marketplaces and refreshes the duplicate groups. It runs in the job tray and the returned groups are from before it. Use for "do I have duplicates?", "find duplicate listings", or before bulk merging.

get_background_job

ChatGPT
Check the status of a background job the assistant started — e.g. a bulk_update_listings re-push, or an inventory/sales import. Pass the jobId those tools returned to see status (pending/running/completed/failed), progress (processed/total/failed/skipped, percent), timing, any error, and results. Omit jobId to list the seller's most recent jobs. Use this to follow up on "did my re-push finish?", to report progress, and — importantly — to VERIFY a bulk edit or import actually completed before telling the seller it's done (don't declare success on the enqueue alone). Read-only.

get_delist_history

ChatGPT
Look up cross-marketplace delist intents (the records SECND creates to pull an item from OTHER marketplaces after it sells on one) for a product or order. Resolve by sku, productId, or orderId. Returns each intent's status (pending/processing/completed/failed), structured failureReason, errorMessage, requiresReconnect flag, attempts, and timestamps. Use this to answer "I sold this item but it didn't get delisted — why?": chain it with get_marketplace_listings to see where the item is still live, then read the delist intents to see whether delisting failed (and why) or was never triggered at all. If a product resolves but has NO intents, that itself is the signal — the sale may not have been detected. Read-only. Scoped to the current seller's own data.

get_delist_results

ChatGPT
Results of a delist_listings or mark_sold job, per product and marketplace: "delisted", "marked_sold", "failed" with the reason and a fix (often reconnecting a marketplace, with a link), or "pending". A listing the marketplace had already ended counts as delisted. Pass waitSeconds (up to 25) to wait. Read-only.

get_listing_results

ChatGPT
Results of a submit_listing job, per product and marketplace: "listed" with the live listing URL, "failed" with the reason in plain words and a fix (reconnect a marketplace with a link, edit the product, resolve something on the marketplace account, or retry later), or "pending" while it is still running. Use this instead of get_background_job after submit_listing. Pass waitSeconds (up to 25) to wait for more items to finish. Read-only.

get_marketplace_listings

ChatGPT
Resolve one or more products (by SKU, title, or product ID — any combination, batched) and return each product's per-marketplace listing records: which marketplaces (Etsy, Grailed, eBay, Shopify, Poshmark, Depop, etc.) it is listed on, plus each listing's ID, URL, listedAt date (when it went up on that marketplace), and SECND's recorded status. This is the primitive for "what platforms is this on?", "here's a list of SKUs — where are they listed?", and (combined with the connected platforms in your SHOP CONTEXT) "what platforms is it NOT on?". Optionally set verifyLive=true to also confirm each listing's status LIVE against the marketplace API (using the seller's connected credentials): each listing then reports its live state alongside the recorded status and a "drift" flag when they disagree — catching items sold or pulled directly on the platform outside SECND. The live value is never masked by the DB value (an un-runnable check reports "unknown"). Read-only: it never delists or changes anything. Exact SKUs/product IDs that match nothing are returned under unmatched (i.e. not in inventory, on no platform). Products the seller DELETED are excluded by default; set includeDeleted: true to surface them (with isDeleted + deletedAt) — use this when the user asks whether an item was deleted, or for a deleted item's history.

get_sell_through

ChatGPT
Show what sells fast and well in the seller's own history against the supply they still have active, grouped by brand, category, subcategory, price band, or marketplace. Per group it returns units sold, revenue, average sale price, median and average days to sell (first listing to sale), sell-through rate (sold / (sold + active supply)), and current active supply. Sellers resell mostly one-of-a-kind items, so use it to answer "what should I restock / source more of?" by pointing at kinds of items (brands, categories, price bands) that sell quickly and well while little supply is left. Also answers "what's selling fastest?", "which brands earn the most?", "which marketplace sells fastest?". Days to sell only counts sales whose list date is known. Active supply is the current count. Read-only, scoped to the seller's own data.

get_slow_movers

ChatGPT
List the seller's active products that have sat longest without selling, oldest first. Each row has productId, a secndUrl deep link, price, days since the product was first listed, each current listing's marketplace and listed date, and a comparison: the seller's own median days to sell for the same brand and the same category over the last year (with the sample size). Use for "which products are moving slowly?", "what's been sitting?", "what should I mark down?". Products with no known list date are excluded and counted in undatedActiveProducts. Views and likes are not tracked. Read-only: it never changes a listing.

get_templates

ChatGPT
Read the seller's current listing templates: title template, description template, and the ordered list of custom instructions attached to each. Returns plain-text content (HTML stripped). Call this BEFORE proposing edits so you know what's already there. Use for "what's in my title template?", "show me my description template", or any rewrite request where you don't already have the current content in context.

lookup_help

ChatGPT
Look up how-to and navigation content from SECND's help knowledge base. Use this when the user asks how, where, or why questions about using SECND itself — finding Settings, connecting a marketplace, how long sync/import takes, understanding a screen, troubleshooting an empty state, or interpreting an error. Returns up to 3 relevant help docs, each with title, body, help deeplink, and (when applicable) an in-app featureDeeplink for navigation. Do NOT use for inventory/sales data questions — those are other tools. Examples: "where is settings?", "how long does sync take?", "how do I connect Etsy?", "where's the restock plan?".

mark_sold

ChatGPT
Record that products sold somewhere SECND does not see (in person, Instagram, a marketplace sale that did not sync) and end every other live listing so nobody else buys them. DRY RUN BY DEFAULT: returns which listings would end; call again with dryRun false after the seller confirms. Needs the seller to allow agents to remove listings. Results: get_delist_results with the jobId.

query_data

ChatGPT
Query the seller's products, orders, or customers from the database. Use this to answer questions about specific products (by brand, category, source/vendor, status, price, tags), orders (by marketplace, date, status, customer), or customers (by name, spending, order count). Supports text search, filtering, and aggregation (count, totals, averages). Examples: "how many products from brand X", "which products have sold", "top customers by spending", "orders from last month". Products the seller DELETED are excluded by default — set filters.deletedOnly (optionally with deletedFrom/deletedTo) to SEARCH deleted products, e.g. "all deleted Marlboro jackets in the past 2 months", or filters.includeDeleted to include both live and deleted. Each product row includes a secndUrl deep link and listedOn: every current marketplace listing with its marketplace and listedAt date. Use listedAt for when an item went up on a marketplace; createdAt is only when it was added to SECND.

recall_memories

ChatGPT
Search what you remember about this seller beyond the memories already in your context: their preferences, rules, business facts, and in-progress workstreams. Matches exact names (brands, marketplaces, suppliers) and paraphrases. Pass includeHistory=true to also see superseded/expired memories and what each current memory replaced — use it for "what did I used to charge?" or "what did I tell you before?". Read-only.

remember_fact

ChatGPT
Save a durable fact, preference, or rule the user has explicitly expressed so it influences future conversations. Examples: "remember I want a $50 minimum margin on jackets", "the supplier for vintage tees is X", "always weight recent weeks more in trend analysis". ONLY use for explicit, durable user statements — NOT for casual chat or single-turn answers. The system already auto-extracts memories from conversation; this tool is for when the user clearly says "remember" or states a long-term rule.

request_photo_uploads

ChatGPT
Get upload slots for photos that exist as LOCAL FILES on the machine you run on (a folder of product photos), for a product the seller owns. Returns, per file, a single-use HTTPS multipart POST (url + fields) valid for 5 minutes: send every field, then the file as the last part named "file"; storage answers 204. Then call attach_uploaded_photos with the uploadKeys that succeeded. Nothing is attached by this tool and no marketplace is touched. Up to 12 photos per call, 20 MB each, JPEG/PNG/WebP/HEIC. For photos that already have a public URL use create_listing_draft.imageUrls instead.

run_batch

ChatGPT
Run up to 25 independent tool calls in one request and get every result back in order. Use it for bulk work (many create_listing_draft, request_photo_uploads or attach_uploaded_photos calls) instead of one request per item. Allowed tools: create_listing_draft, request_photo_uploads, attach_uploaded_photos, query_data, suggest_pricing, get_listing_results, get_delist_results, get_background_job, search, fetch. Calls cannot reference each other's results; run dependent steps as separate batches (all drafts, then all upload slots, then all attaches). Publishing, price changes and removals are not batchable. Each call is checked and logged exactly as if called directly; a refused or failed call does not stop the others unless stopOnError is true.

search_by_image

ChatGPT
Find products in the SELLER'S OWN inventory that visually match a given image, using image-embedding similarity (pgvector). Pass the URL of an image (e.g. one the user attached in chat, or a product photo) and it returns the closest-matching products with a similarity score. Use for "find this jacket in my inventory", "do I already have something like this?", "which of my listings matches this photo?". Requires the seller's catalog to be embedded. Read-only, scoped to the seller's own products.

show_products

ChatGPT
Show specific products to the seller as product cards with photos, price, status, and the marketplaces each is live on, plus a "Review all" action when the products can be reviewed. Pass productId values from query_data, get_slow_movers, or get_marketplace_listings (at most 24). The server looks up every field itself and refuses ids that are not the seller's, so never type product details yourself. The result includes a fence string: paste it on its own lines exactly where the cards should appear in your reply. In MCP clients the result also carries small product photos.

submit_listing

ChatGPT
Publish products to marketplaces through SECND's submission pipeline (cross-listing). Takes productIds (drafts from create_listing_draft or any existing products) and target marketplaces the seller has connected. Runs readiness preflight first — if products are missing required marketplace fields, the response lists exactly what to fix instead of submitting. Expired marketplace connections are refused with a reconnect link. Pass waitSeconds (up to 25) to wait for results in the same call; otherwise call get_listing_results with the returned jobId. Results say, per product and marketplace, whether it is live (with its URL) or failed (with the reason and what to do). ALWAYS confirm the products and marketplaces with the user before calling — this publishes live listings on external marketplaces.

suggest_pricing

ChatGPT
Suggest a price for a product based on comparable sold listings across marketplaces. Returns a recommended price, range (low/high), and the comparables it pulled from. Use this for "what should I price this at?", "is my $40 price right?", or "what are similar items selling for?". Best for individual items — pass a descriptive title.

update_prices

ChatGPT
Change product prices in SECND and, by default, push the new price to every marketplace where each product is live (existing listings are updated in place, not relisted). DRY RUN BY DEFAULT: returns current vs new price, the % change, where it would go live, and warnings (e.g. a drop over 50%); call again with dryRun false after the seller confirms. Push results: get_listing_results with the returned push jobId.

update_template

ChatGPT
Write the seller's title OR description template, and optionally replace its custom instructions. Content is stored as plain text wrapped in a paragraph — placeholder chips (e.g. @measurements) are NOT created by this tool; the user adds those manually. Pass instructions as the FULL replacement list (omit to leave instructions untouched). Use when the user asks to rewrite, shorten, expand, or otherwise change their template. ALWAYS call get_templates first if you don't already have the current content.

attach_uploaded_photos

Claude

bulk_update_listings

Claude

create_listing_draft

Claude

delist_listings

Claude

fetch

Claude

find_duplicate_products

Claude

get_background_job

Claude

get_delist_history

Claude

get_delist_results

Claude

get_listing_results

Claude

get_marketplace_listings

Claude

get_sell_through

Claude

get_slow_movers

Claude

get_templates

Claude

lookup_help

Claude

mark_sold

Claude

query_data

Claude

recall_memories

Claude

remember_fact

Claude

request_photo_uploads

Claude

run_batch

Claude

search_by_image

Claude

show_products

Claude

submit_listing

Claude

suggest_pricing

Claude

update_prices

Claude

update_template

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/secnd"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/secnd)