A secure bridge that lets the team ask Claude plain-English questions about our LoudCrowd creator sales — revenue, orders, products, commission owed per creator, and a per-influencer leaderboard — with complete history kept for instant analysis.
loudcrowd-mcp.melindamaria.ai/mcp. Reads LoudCrowd-attributed orders from Shopify, mirrors all 10,229 orders in a Cloudflare D1 database, refreshes nightly, and answers any range instantly — including the all-time influencer leaderboard. Update (Aug 1): commission is LIVE. The connector now syncs per-order commission dollars + attribution types nightly from LoudCrowd's official Brand API (released Jul 28), across both programs (Ambassadors + Creators): 2 new commission tools, and commission built into the summary and leaderboard. The connector is fully built for every data feed LoudCrowd has shipped; engagement & member data arrive when LoudCrowd releases those endpoints — see Limitations.The v5 upgrade (Aug 1) replaced the old connector in place — same connector, same sign-in, nothing new to add. But Claude only refreshes a connector's tool list when you reconnect, so until you do, the two new commission tools stay invisible to you.
Connecting for the first time instead? Skip this — you'll get v5 automatically. Full steps in How to connect below.
LoudCrowd is Melinda Maria's creator-commerce platform — it runs our ambassador, affiliate, and UGC programs and drives a meaningful slice of sales. (A "creator/ambassador platform" = the system that gives influencers tracked links and codes, then attributes the sales they drive.)
Until now, answering "which influencers are selling the most?" meant logging into LoudCrowd or waiting on their reporting. The connector removes that friction: anyone on the team can just ask Claude — "who are our top LoudCrowd creators this month?", "what did LoudCrowd drive last quarter?", "what's selling through our ambassadors?" — and get an instant, accurate answer.
It works by reading the orders LoudCrowd tags in Shopify (LoudCrowd stamps each ambassador order with the creator's handle), keeps a complete local copy so trend questions are instant, and never depends on a slow export.
A real influencer leaderboard by creator handle — revenue and orders per person, ranked, for any time period.
Total revenue, order counts, and average order value — last 7 / 30 / 90 days, quarter-to-date, or all-time.
The exact products and SKUs moving through LoudCrowd, ranked by units — so you see the hero pieces.
Its own cached database means even "all-time" answers come back in a second — no waiting, no timeouts.
🔄 Connected before August 1, 2026? The v5 commission upgrade replaced the old connector in place — disconnect and reconnect once (Settings → Connectors → LoudCrowd) so Claude refreshes the tool list and picks up the two new commission tools. See the pink box at the top of this page for the 3-step version.
In Claude (desktop or web), open Settings and go to Connectors. LoudCrowd is already loaded for the whole team — there is nothing to add and no URL to paste.
Scroll to LoudCrowd in the connector list and click Connect. (Connected before Aug 1, 2026? Click Disconnect first, then Connect — that one-time refresh unlocks the v5 commission tools.)
A Google sign-in opens. Choose your @melindamaria.com (or @7ate9.com) account. Any other domain is rejected. You never type a Shopify token or password.
Approve the connection. To confirm it works, ask Claude to "run loudcrowd_health." Then ask anything from the use cases below — in plain English. Just include the word LoudCrowd.
You will never enter a Shopify token, API key, or any credential into chat. Sign-in is your own Google account only. The connector holds the Shopify key internally and uses it on your behalf — read-only.
The connector is read-only by design. Commission dollars now come from LoudCrowd's official Brand API (synced nightly, coverage from ~Apr 2026). Engagement metrics and member counts still live only inside LoudCrowd until they ship those endpoints — see Limitations.
Who's driving the most sales — revenue & orders per creator by their real handle, ranked, for any period. Unattributed sales reported separately and honestly.
Total revenue, order count, average order value, paid vs. $0 comp orders — last 7 / 30 / 90 days, quarter-to-date, all-time, or custom. Now includes the period's commission total automatically.
What MM owes creators — per period, per creator, or per order — straight from LoudCrowd's official Brand API, with attribution types (tracked visit, link discount, etc.) and refund-adjusted amounts. Ask "how much commission do we owe this month, by creator?"
This month vs. last (or any 7/30/90-day window vs. the one before), with % change on revenue, orders, and active creators.
Which products sold the most units through LoudCrowd — overall, or for one specific creator ("what does maymaytoldme sell best?").
Every LoudCrowd order with date, value, channel, creator, and code — viewable, or dumped as rows for a spreadsheet.
Backed by a local database, even all-time/lifetime questions answer in a second — no waiting, no timeouts.
You don't need to memorize tool names — just ask Claude in plain English and it picks the right one. This catalog is here so you know the full breadth. Everything is read-only.
loudcrowd_summary (revenue / orders / AOV / paid-vs-comp) · loudcrowd_top_creators (the influencer leaderboard, by handle + LoudCrowd id) · loudcrowd_compare (this period vs. previous, % change) · loudcrowd_sku_breakdown (top products by units) · loudcrowd_creator_products (one creator's best SKUs)
loudcrowd_commission (the commission report: totals, per-creator leaderboard, attribution + program breakdowns, refunds — for any period) · loudcrowd_order_commission (one order's official record: commission, attribution types, program, refund amounts)
loudcrowd_orders (individual orders: date, value, channel, code) · loudcrowd_export (raw rows for CSV / Sheets: order, date, revenue, creator, code, channel, SKUs)
loudcrowd_health (is it reading Shopify? how many tagged orders?) · loudcrowd_health_detail (exact query + page count, if a report looks off)
A nightly sync (Cloudflare Cron) refreshes the database automatically — the day's new/changed Shopify orders, then the commission sync from LoudCrowd's official Brand API (both programs, refund-adjusted). Every result is tagged source:"d1" (from the cache) or source:"shopify" (live fallback) so you always know where a number came from.
It's a powerful tool, but a few things are worth knowing so you trust the numbers. Here they are in plain terms.
The connector can answer questions and run reports, but it cannot change anything — it can't edit, cancel, or refund an order, or change anything in Shopify or LoudCrowd. That's on purpose, so it's completely safe to use.
For a sale to show a creator's name, LoudCrowd has to record that name on the order — and they only started doing this recently. So recent sales are well covered (about 8 in 10 name the creator), but older ones often aren't (about 4 in 10). The sales without a name still count in every revenue total — they just show up in an "unknown creator" group on the leaderboard, because the name was never recorded and can't be invented after the fact.
Across all time, about $83,500 of real sales has no creator name attached. Some are simply old (from before names were recorded); others were credited to a different program (like ShopMy or Impact) instead. This isn't an error — the connector always shows this "unknown" amount on its own so the totals stay honest.
Because the connector keeps its own copy of every order, even "all-time" questions answer in a second. There's a backup mode that reads live from Shopify if that copy is ever unavailable — and that backup only goes back about 3,000 orders. In normal day-to-day use you'll never hit this.
The connector updates its copy of the data every night (around 2am). So a sale from this afternoon shows up in totals tomorrow morning. If you need this-minute numbers, just ask for the "latest" or "live" figures and it checks Shopify directly.
Update (Aug 1): commission dollars are now in the connector, synced nightly from LoudCrowd's official Brand API. One honest caveat: LoudCrowd's API only covers orders since our programs were created (~April 2026), so older orders in the lifetime mirror have no commission figures — that history simply doesn't exist at LoudCrowd's end. Social numbers (posts, impressions, sessions, ROAS) and member counts are still LoudCrowd-only until they ship those endpoints — the connector says so honestly rather than showing zeros, and they'll plug straight in when LoudCrowd releases them (nothing we built changes).
Live-verified against the mirror: official attribution (commission $ + how the sale happened) exists only from April 2, 2026 — the day our programs were created; LoudCrowd's API has no earlier history. The link-vs-code channel split also only exists from around then: 98.9% of the 10,235 lifetime orders (June 2024 → today) carry no channel tag. Creator identity is better: 41% of lifetime orders carry a creator handle stamped at checkout — that's what powers the all-time leaderboard — but 59% (6,033 orders) have no creator or channel data at all. Plain rule: revenue totals are solid for any period; commission, channel-mix, and complete per-creator reports are only trustworthy from April 2026 onward. Every loudcrowd_summary answer now carries an attribution_coverage block saying exactly this.
The connector trusts how LoudCrowd labels each order. If an order is mislabeled or missed, the report reflects that. Update (Jun 30): the empty "discount code" tag is not a gap — our ambassador program doesn't use affiliate codes, and LoudCrowd confirmed Shopify link-tagging already captures every LoudCrowd order. For anything high-stakes, still double-check against the source.
The connector is complete for every data feed LoudCrowd offers today — revenue, orders, products, the leaderboard, and (since Aug 1) official commission across both programs. Two more feeds are on LoudCrowd's roadmap; because the connector is a ready-made API client, each one plugs into the same nightly sync when it ships — no rebuild, nothing we built is wasted.
Impressions, posts, reach, sessions, conversion rate, and ROAS will ship as a separate LoudCrowd endpoint (their timeline). Until then the tools say "not yet available from LoudCrowd" — never fake zeros.
Total members, new signups, active members (who posted), and purchasing members arrive as another endpoint after engagement.
The LoudCrowd-Code tag is empty simply because our ambassador program doesn't use affiliate codes — LoudCrowd confirmed that Shopify link-tagging already captures every LoudCrowd-driven order. Codes aren't needed; this isn't a gap; no action required.
Everything measurable today is measured: revenue, orders, products, the leaderboard, and commission. The two feeds above are LoudCrowd's to release — and the day they do, they slot straight in.
The real power shows up when a question spans tools — Claude pulls from each and stitches the answer together. These combine LoudCrowd with Shopify (DTC store), ShopMy (creator/affiliate sales), Klaviyo (email), Slack, Asana, Gmail, and Canva. You must have the relevant connector added. Tip: name both systems so Claude knows where to look.
These go further: each one chains LoudCrowd's official commission & program data (new in v5) with several other connectors into a full workflow. Copy any of them straight into Claude — and name every system, like these do, so Claude knows where to look. Anything outbound (email, Slack, tasks) is drafted for your approval first.
The cross-connector examples only work if you've added the other connector (Shopify, ShopMy, Klaviyo…) to your Claude. Matching records across systems relies on shared identifiers (creator handles, codes) being consistent — where they differ, Claude flags the ambiguity rather than guesses.
When a prompt involves messaging (Slack, Gmail DMs to creators), Claude prepares a draft for your review — it won't send to people on its own. You approve before anything goes out.
Revenue figures are summed from Shopify order data. For official accounting, treat Shopify (or the accounting system) as the source of truth — the connector reports attributed order value, not booked revenue.
Only @melindamaria.com and @7ate9.com Google accounts can connect. No public access. If someone leaves, removing their Google access cuts off the connector too.
Read-only access · MM-staff-only login · the Shopify secret is never exposed · and the local data copy stays inside Melinda Maria's own Cloudflare account. Same hardened pattern as our other production connectors. None of the residual risks below are rated high.
The connector cannot write, so a bug or misuse can't alter, cancel, or corrupt anything in Shopify or LoudCrowd. Worst case is reading data.
Mitigation: no write tools exist. Scopes are limited to read_orders + read_products.
The Shopify Admin credential is powerful. If leaked, someone could read order data.
Mitigation: stored only as encrypted Cloudflare secrets (Client ID + Secret), used server-side, never sent to the browser or chat. The access token auto-expires every 24h and refreshes. Rotate in the Shopify Dev Dashboard anytime — the old one dies instantly.
Someone outside MM trying to connect and read company sales data.
Mitigation: Google OAuth with a strict email-domain allow-list (@melindamaria.com / @7ate9.com). Other domains are rejected at login. OAuth 2.1; encrypted session cookies.
There are no per-person limits inside the connector — any approved MM/7ate9 user can view the full leaderboard, revenue, and order export. Fine for an internal tool, but worth knowing before sharing widely.
Mitigation: access is gated to two company domains; treat connector access like access to a sales spreadsheet. Remove Google access when someone leaves. Consider limiting who you share the URL with.
A complete copy of LoudCrowd order data — amounts, products, creator handles, discount codes — lives in a Cloudflare D1 database.
Mitigation: D1 sits inside MM's own Cloudflare account, reachable only through the authenticated connector (no public endpoint). It holds order facts only — no payment details, no full customer profiles. Treat the mirror as confidential company data; document who can access the Cloudflare account.
If the nightly sync stops (token revoked, error), the mirror could quietly fall behind without anyone noticing.
Mitigation: each result is tagged source, and reports fall back to live Shopify if the cache is stale/empty, so answers stay correct. Recommended add: a freshness/uptime alert (we have the pattern from other connectors) so a stalled sync emails the owner.
Text in an order field (a note, a discount code) could in theory try to instruct Claude.
Mitigation: tools are read-only, so injected text can't trigger a write. Claude treats tool output as data, not commands. Low impact in a read-only design.
Heavy live querying could hit Shopify's API limits and slow responses.
Mitigation: the D1 mirror absorbs nearly all reads, a 30-min KV cache covers repeats, and live pulls retry with backoff and a page cap.
A live artifact is a small interactive page Claude builds and saves for you — a dashboard, tracker, or report that pulls fresh data from this connector every time you open it, instead of being a frozen snapshot. In a Cowork session, just ask: “create a live artifact that…” — it lands in your Artifacts list, reusable forever, and its charts and tables refresh themselves on open.
Sharing with the team: artifacts are saved to your Claude account. Three easy ways to share: (1) ask Claude to export a snapshot (PDF, image, or .xlsx) and drop it in Slack; (2) send a teammate the prompt that built it — their copy fills with live data through their own connector sign-in; (3) for a permanent all-team dashboard, that’s the MM Intelligence project on the AI roadmap.
The connector is a single Cloudflare Worker wearing four hats (v5): an OAuth 2.1 server (who can use it), an MCP server (what Claude calls), a Shopify reader (revenue backbone), and a LoudCrowd Brand-API reader (official commission + attribution, both programs) — backed by a KV cache and a D1 database, with a nightly cron that syncs orders first, then commission.
tools/call as JSON-RPC over HTTPS/mcp · the connector's brainLoudCrowdtools/call over HTTPS.getOrders() hits D1 first — a fast SQL read for the date range.source.scheduled() at ~2am — no laptop, no browser.last_sync from D1's lc_meta table.updated_at ≥ last_sync — just the daily delta.lc_orders — without ever touching the v5 commission columns.last_sync. A missed night auto-catches-up next run.lc_commissions, and stamps the matching mirror rows. Each step has its own safety net, so one failing never blocks the other.One reusable skeleton. Same proven pattern as our ShopMy & other connectors: a Cloudflare Worker that is its own login server, with one "client" module that knows how to read Shopify. The Shopify secret lives only inside the Worker — never in chat, never shown to users. One-time setup, not in the live path: a backfill.mjs script ran once on a Mac to seed all 10,141 historical orders (batched at 50/statement to stay under SQLite's size limit).
All the history lives in one database (Cloudflare D1 — a small SQL database inside MM's own Cloudflare account) with two layers: a lifetime revenue backbone captured from Shopify, and the official commission layer captured from LoudCrowd's Brand API. The picture below is drawn to scale — see how far back each layer reaches, and why both exist.
A one-time backfill script paged through Shopify's full order history to seed the backbone. Since then, a scheduled job (a "cron" — an automatic timer in Cloudflare's cloud, no computer needed) runs nightly at ~2am and adds only that day's new or changed orders. A watermark remembers where it left off, so a missed night self-heals.
Right after the Shopify step, the same nightly run re-reads LoudCrowd's official API end-to-end — it's small, so it takes seconds. That full re-read is deliberate: if a refund months later claws back a creator's commission, we catch it. Each API order is matched to its backbone row by Shopify order ID.
Ask Claude anything — tools read the database first, so even lifetime questions answer in about a second. If the mirror were ever unavailable, a live-Shopify backup covers the newest ~3,000 orders. Every result names its source, so the numbers are always traceable.
Each layer sees things the other can't. The backbone holds ~10,000 orders from before April 2026 that LoudCrowd's API will never return. The commission layer attributes ~140 program-era orders that carry no LoudCrowd tag in Shopify — invisible to the backbone. Keeping both, joined by order ID, is what makes the connector's answers complete: lifetime revenue from one, official payout truth from the other.