> Illustrative sample produced with the Cheek Company Knowledge Graph skill. Mastercard is not affiliated with and has not endorsed this material. Page counts, chunk counts, and graph figures are plausible sample values, not the output of a real crawl.

# Vault Summary: mastercard.com

This is the portable summary of the local intelligence vault built for mastercard.com. The vault itself lives on disk at `companies/mastercard.com/`; this file is what sibling skills read when they need the company without re-researching it.

## What this vault is

A local, continuously refreshed archive of everything Mastercard says in public: the website and its subdomains converted to Markdown, the latest SEC filing converted to Markdown, and the owner's own uploads, all embedded for semantic search, mapped as a knowledge graph, served by an MCP server and an always-on dashboard, and refreshed daily by cron with a committed, diffed, summarized record of what changed.

- **Company:** Mastercard Incorporated, NYSE: MA, SEC CIK 0001141391
- **Registrable domain:** mastercard.com
- **Organization type:** public company
- **Built:** August 12, 2026
- **Rendering engine:** Playwright. The main site is script-rendered, so pages are rendered in headless Chromium before conversion, with changed-pages-only screenshots enabled.
- **Crawl cap:** 2,000 pages (the default)

## What was crawled

1,842 pages captured as Markdown under `sources/web/<host>/`, mirroring each URL path. Counts per host:

| Host | Pages | Notes |
|---|---|---|
| www.mastercard.com | 1,214 | Corporate site, newsroom under /news, product and vision pages |
| investor.mastercard.com | 236 | Investor relations: releases, events, governance documents |
| developer.mastercard.com | 189 | Truncated by the 2,000 page cap; the developer portal is larger than the remaining budget. Raise the cap to complete it. |
| b2b.mastercard.com | 87 | Business solutions pages |
| careers.mastercard.com | 41 | Landing and program pages only; the Workday job listings render behind search interactions the crawler does not submit. Use the Job Postings Scan skill for the postings themselves. |
| other subdomains | 75 | Brand center, regional marketing hosts discovered from links |

Politeness settings used: robots.txt respected, default request delay and concurrency, honest user agent carrying the owner's contact email, crawl confined to mastercard.com and its subdomains.

## Filing captured

- **FY2024 Form 10-K**, filed February 2025, fetched from SEC EDGAR by CIK 0001141391.
- Original document at `sources/filings/mastercard-10k-2024.htm`, converted Markdown beside it at `sources/filings/mastercard-10k-2024.md`.
- Headline figures for quick orientation: net revenue $28.2 billion, net income $12.9 billion, approximately 33,900 employees, more than 200 countries and territories.
- Uploads folder `sources/uploads/` is live; anything placed there joins the index on the next refresh.

## Index and graph shape

- **Embeddings:** 36,510 chunks from 1,843 source documents (web pages plus the converted 10-K), embedded with the local model into embedded Postgres with pgvector at `db/`, with a read index exported for the dashboard and the MCP server.
- **Knowledge graph:** 2,317 nodes and 9,408 edges in `graphiffy/`. Node types: pages, organizations (Mastercard, acquired companies such as Recorded Future and Dynamic Yield, partners), people (the leadership roster from the site's leadership page), products (Decision Intelligence, Agent Pay, Shopping Muse, Test and Learn, Consumer Fraud Risk), and places (technology hubs including O'Fallon, Dublin, Pune, Vancouver). Edge types: links-to, mentions, part-of.
- **Git:** the `sources/` tree is a git repository; the initial commit captures the full first crawl plus the filing.

## How to search it

- **MCP server:** registered with the Claude CLI as `vault-mastercard-com`. Ask any Claude session semantic questions against the vault, for example: "what does Mastercard say about agentic commerce," "which services products mention fraud," "what changed on the investor site this week."
- **Dashboard:** always-on at http://localhost:4310 with three views: the interactive knowledge graph, semantic search, and the daily change log.
- **On disk:** every page is plain Markdown under `sources/web/`, so grep works too.

## How the refresh works

A daily cron job at 06:15 local time rescrapes the domain with the same deterministic rules (same site in, same files out, no timestamps in generated files), re-embeds anything that changed, commits the result, writes a full diff file for the day, and summarizes the diff with the Claude CLI into the change log the dashboard shows. Watch the change log for new pages under /news and investor.mastercard.com; those two hosts carry most day-to-day movement.

## What is not in the vault

Honesty section. The developer portal is incomplete at the current cap. The Workday postings are not captured here by design. Anything behind login, and any page robots.txt disallows, is absent. No non-public information exists anywhere in this vault; private material enters only through the uploads folder on the owner's machine.

IMPORTANT: download this file and upload it to your Claude project, or keep it in this working folder if you are in Claude Code, so the other skills in this collection can find it and build on your work. The natural next step is the AIDE Opportunities Audit (/cheek-aide-opportunities-audit), which reads this vault instead of re-researching the company.