weekly-meeting

name	weekly-meeting
description	Prepare weekly meeting data — fetches 30-day Marketplace revenue, new/large/lost customers, and updates the "Weekly Meeting Prep" panel at the top of the client profiles homepage (http://localhost:52401/). Use when the user says "weekly meeting", "prep for the meeting", "update the weekly meeting section", or "what's our revenue this week".

Weekly Meeting Prep

Fetches Atlassian Marketplace transaction data for the past 30 days, processes it, and updates the "Weekly Meeting Prep" section in private/client-profiles/index.html.

Prerequisites

A browser tab must be open and logged into marketplace.atlassian.com (as the vendor). The skill uses that session's cookies via mcp__playwright__browser_evaluate to call the REST API — no separate auth token is needed.

If the browser is not already on a Marketplace page, navigate there first:

• Target page: https://marketplace.atlassian.com/manage/vendors/1215266/reporting/sales?period=past30Days

Step 1 — Fetch all transactions (past 30d)

Use mcp__playwright__browser_evaluate to call the REST API from the browser context (authenticated via cookies). Paginate until all transactions are collected.

async () => {
  const today = new Date();
  const endDate = today.toISOString().split('T')[0];
  const start = new Date(today);
  start.setDate(start.getDate() - 30);
  const startDate = start.toISOString().split('T')[0];

  let allTx = [];
  let offset = 0;
  const limit = 100;

  while (true) {
    const url = `https://marketplace.atlassian.com/rest/2/vendors/1215266/reporting/sales/transactions` +
      `?endDate=${endDate}&startDate=${startDate}&excludeZeroTransactions=true` +
      `&limit=${limit}&period=past30Days&offset=${offset}`;

    const resp = await fetch(url, { credentials: 'include' });
    if (!resp.ok) return JSON.stringify({ error: resp.status });
    const data = await resp.json();
    const txs = data.transactions || [];
    allTx = allTx.concat(txs);
    if (txs.length < limit) break;
    offset += limit;
  }

  // Process
  const revenue = { total: 0, vendor: 0 };
  const byType = {};
  const byCompany = {};

  for (const tx of allTx) {
    const pd = tx.purchaseDetails;
    const cd = tx.customerDetails;
    revenue.total += pd.purchasePrice || 0;
    revenue.vendor += pd.vendorAmount || 0;
    byType[pd.saleType] = (byType[pd.saleType] || 0) + 1;

    const key = cd.company;
    if (!byCompany[key]) {
      byCompany[key] = { total: 0, vendor: 0, types: [], country: cd.country };
    }
    byCompany[key].total += pd.purchasePrice || 0;
    byCompany[key].vendor += pd.vendorAmount || 0;
    byCompany[key].types.push(pd.saleType);
  }

  const companies = Object.entries(byCompany);
  const r = v => Math.round(v * 100) / 100;

  return JSON.stringify({
    startDate, endDate,
    totalTransactions: allTx.length,
    totalRevenue: r(revenue.total),
    vendorRevenue: r(revenue.vendor),
    byType,
    newCustomers: companies
      .filter(([, v]) => v.types.includes('New'))
      .map(([k, v]) => ({ company: k, country: v.country, total: r(v.total) })),
    upgradedCustomers: companies
      .filter(([, v]) => v.types.includes('Upgrade'))
      .map(([k, v]) => ({ company: k, country: v.country, total: r(v.total) })),
    downgradedCustomers: companies
      .filter(([, v]) => v.types.includes('Downgrade'))
      .map(([k, v]) => ({ company: k, country: v.country, total: r(v.total) })),
    largestCustomers: companies
      .sort((a, b) => b[1].total - a[1].total)
      .slice(0, 5)
      .map(([k, v]) => ({ company: k, country: v.country, total: r(v.total), types: v.types }))
  });
}

Parse the returned JSON to extract the data fields.

Step 2 — Build the Weekly Meeting HTML section

Using the data from Step 1, build the replacement HTML block. The section is identified by the comment <!-- Weekly Meeting Prep — updated by /weekly-meeting skill --> in index.html.

Revenue KPI strip: 5 tiles — Revenue (30d), Vendor take (30d), Renewals count, New customers count, Largest deal.

Three-column grid:

• New customers: List companies with saleType="New". If none, show "None this period" in muted text.
• Largest customers (30d): Top 5 by total purchase price. Color downgraded companies red (#c0392b) and annotate ↓, color upgraded green (#16a34a) and annotate ↑.
• Upgraded / Downgraded: List saleType="Upgrade" entries in green, saleType="Downgrade" in red.

Date range label: {startDate} → {endDate} · Updated {today}.

Step 3 — Update index.html

Find the block between:

<!-- Weekly Meeting Prep — updated by /weekly-meeting skill -->

and the next:

      <div class="hp-section">
        <h2>All profiled tenants</h2>

Replace the entire block (from the comment down to, but not including, the "All profiled tenants" section) with the newly generated HTML.

Edit private/client-profiles/index.html using the Edit tool.

Step 4 — Commit and surface

cd ~/workspaces/zenuml/conf-app/private
git add client-profiles/index.html
git commit -m "chore(client-profiles): weekly meeting prep $(date +%Y-%m-%d)"

Then bump submodule pointer:

cd ~/workspaces/zenuml/conf-app
git add private
git commit -m "chore(private): bump submodule — weekly meeting $(date +%Y-%m-%d)"

Ensure server is running and print:

→ http://localhost:52401/#weekly-meeting

Key facts

• Vendor ID: 1215266
• REST API base: https://marketplace.atlassian.com/rest/2/vendors/1215266/reporting/sales/transactions
• Auth: browser cookies (no API key needed when fetched from mcp__playwright__browser_evaluate while logged in)
• Sale types: New, Renewal, Upgrade, Downgrade, Refund
• "New customers" = saleType: "New" — first-ever paid subscription
• "Large customers" = top 5 by purchasePrice in the 30d window
• "Lost/at-risk" = saleType: "Downgrade" or changeInTier: "Decrease" in Renewal transactions
• Server port: 52401
• Output file: private/client-profiles/index.html