What it does
The Apex Flow 3D Scraper extracts public pages as clean Markdown, with CSS selector field extraction, whole-site crawls of up to 1,000 pages run as background jobs, and parallel batches of up to 20 pages on the Scale plan. Each read reports whether a page is new, unchanged or changed, and can return only the diff. Pages can be split to fit a model's context window, and when a live site refuses a request, a dated Internet Archive copy can be returned instead. The publisher's description also mentions house-price and news data tools.
It suits developers and AI agents that need web content inside an assistant such as Claude Code, Codex or VS Code, or that want to monitor pages for changes. The MCP endpoint is listed in the official MCP Registry, and Python and Node clients are included. The publisher also provides a separate, MIT-licensed engine that can be self-hosted, with a CLI, durable SQLite jobs, JSON and CSV exports and search across saved results.
Using the hosted service requires a free API key, which includes 1,000 credits, and a bearer key for Streamable HTTP MCP clients; a plain page costs one credit and a rendered page costs three. The self-hosted engine requires Python 3.11 or later, and Brave must be installed separately for JavaScript rendering. HTTP-only collection does not need a browser. A local workspace runs on loopback for one machine.
Limitations and requirements
The hosted service requires a free API key and a bearer key, and credits are limited; failed and blocked pages cost nothing, and there are no automatic overage charges. Self-hosting requires Python 3.11 or later, and JavaScript rendering requires a separately installed Brave browser. Run one worker per data directory. The local web interface is not a public multi-tenant hosting service and has no account or authorization system for internet exposure. An optional proxy is only a proxy you control; it does not provide protected-site access. The publisher states that compute, bandwidth, storage and maintenance are not free, and that collected page content is untrusted data that an assistant must not execute.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 14 Nov 2026 · its page on the source
Apex Flow 3D Scraper
Scrape, crawl and monitor public pages as Markdown, with house-price and news data tools.
v0.10.0As shown on registry.modelcontextprotocol.io when we last read it