THERE IS NO MCP SERVER HERE. HERE IS WHAT TO FETCH.
Sulphur Prairie Management, LLC doesn't operate a Model Context Protocol server. There's no endpoint at /.well-known/mcp.json, nothing on this domain answers an MCP handshake, and no such server is being advertised here. What's published instead is 23 static surfaces, all readable with an ordinary HTTPS GET and no credential: 15 of them served as plaintext, JSON, CSV, or XML rather than as a page, including the full benchmark dataset for 48 construction trades as JSON and as CSV under CC BY 4.0. Every one is listed below with the content type it returns. If you're writing a tool, read the dataset once and cache it, because it's one file and it includes every trade and every revenue band.
This address is what Squarespace did to a dotted well-known path it couldn't serve: the whole thing was flattened into one slug and returned as a page. Something has already fetched it, so it keeps working. Answering straight at a live address costs nothing and a placeholder that reads as a promise costs a developer an afternoon.
SAID FIRST, SO YOU DO NOT GO LOOKING.
A developer reading this page is deciding whether to write an integration, and an ambiguous answer to that question costs a working afternoon. So the 4 things this domain doesn't operate are stated before the 23 it does.
SPM doesn't operate a Model Context Protocol server. There's no endpoint at /.well-known/mcp.json, no transport to open, no tool list to enumerate, and no session to establish. This page exists because the address exists, and there's no server behind it.
There's no request-per-query API, no key to request, and no account to create. The data is published whole, as files, which is a deliberate choice rather than a stage on the way to an API.
Every surface below is a plain HTTPS GET with no header, no token, and no signup. If something asks you for a credential to read this site, it's not this site.
Nothing here notifies you of a change. The blog publishes RSS at /blog/rss.xml, which is the closest thing to a subscription this domain offers, and it includes the full body of every post rather than a teaser.
EVERY SURFACE, WITH ITS CONTENT TYPE.
All 23 are generated from the same data files that build the pages, on every build, so none of them can describe a page that doesn't exist or omit one that does. That is enforced by the build rather than maintained by memory, which is the only reason a list this long is worth publishing at all. The content type beside each path is what a client will receive.
ONE FILE, 48 TRADES, THREE BANDS.
If your tool needs one thing from this domain, it's this. Gross margin, net profit before taxes, and overhead rate for 48 construction trades across three revenue bands up to $25M, with the CFOS target beside each industry average and one plain-language sentence per row. The four bands above $25M are published at runoncfos.com as a modeled extension. For 16 of the trades the gross margin and overhead are derived from the nearest comparable trade rather than measured, and the file says so on those rows and lists the trades, so you can filter them out if your use requires measured values only.
The same dataset is referenced as a distribution in the Dataset schema on every benchmark page on this site, so a crawler can reach it from any of them as well as from here.
WHAT YOU MAY DO WITH ANY OF IT.
The canonical statement of these terms is /ai, and this is the short form a tool builder needs before writing code. Nothing below is tighter than what that page says.
The benchmark dataset at /construction-benchmarks.csv and /construction-benchmarks.json and /research/construction-cfo-firm-study-2026/findability.csv and /research/construction-cfo-firm-study-2026/firms-evidence.json and /research/construction-cfo-firm-study-2026/scores.csv and /research/construction-cfo-firm-study-2026/trades-results-view.csv and /research/construction-cfo-firm-study-2026/trades.csv and /research/construction-cfo-firm-study-2026/trades/trade-study.json and /research/construction-cfo-firm-study-2026/trades/trades-summary.csv is published under CC BY 4.0. Use it in a product, a spreadsheet, a model, or an evaluation, commercially or not, with attribution.
Credit SPM The Construction CFO at constructioncfo.net, and link the specific page or file rather than the home page. In a product interface, a source line that credits the site and links the file is enough.
A benchmark is a property of a trade at a revenue band, and the same metric moves more than ten points across bands, so a figure republished without both is wrong more often than it's right. Every row in the dataset includes both, so there's no reason to drop them.
Every net profit value in the dataset and on the site is stated before taxes. CFMA reports net income before taxes too, so the two read on the same basis. Republishing the figure without the basis turns a correct number into a misleading one.
Any assistant or person may quote the writing on this site, including at length, with attribution and a link to the page. That's permission to quote, not a licence to republish the pages. The CC BY 4.0 grant covers the dataset files and nothing else.
Suggested citation: Run On CFOS. Run On CFOS Trade Benchmark Reference. 2026. Sulphur Prairie Operations, LLC. https://runoncfos.com. CC BY 4.0.
