About OwnerSpec
OwnerSpec is an MCP server for cited guidance on home water treatment, plumbing, and air equipment, aimed at AI agents that need verifiable answers. Responses include citation URLs and fact-verification dates.
Key features
- 01Search guides, calculators, parts, and reviews
- 02Fetch sourced pages and quick answers
- 03Interpret water reports and diagnose issues
- 04Match replacement parts and product picks
- 05Run calculators and plan maintenance
- 06Convert water hardness and size softeners
Use cases
- 01Compare test results with EPA limits
- 02Find compatible filters or lamps by model
- 03Estimate softener capacity and salt efficiency
- 04Create a sourced home maintenance schedule
MCP support
It exposes read-only tools, resources, and the answer_from_ownerspec prompt to MCP clients. Some tools render interactive views in hosts that support MCP Apps.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://ownerspec.com/mcpclaude mcp add --transport http com-ownerspec https://ownerspec.com/mcp{"mcpServers":{"com-ownerspec":{"type":"http","url":"https://ownerspec.com/mcp"}}}Open in Cursor{"servers":{"com-ownerspec":{"type":"http","url":"https://ownerspec.com/mcp"}}}Open in VS Code{"mcpServers":{"com-ownerspec":{"type":"http","url":"https://ownerspec.com/mcp"}}}{"mcpServers":{"com-ownerspec":{"serverUrl":"https://ownerspec.com/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
8| Parameter | Type | Required | Description |
|---|---|---|---|
| layer | enum | No | Restrict to one layer: guides (diagnose), calculators (size), parts (match a model to a part), reviews (compare products)guides | calculators | parts | reviews |
| limit | integer | No | Max results (default 5) |
| query | string | Yes | Search terms, e.g. "iron well water orange stains" or "RCC7 filters" |
| topic | string | No | Optional topic slug such as softening, filtration, iron, reverse-osmosis, uv, testing |
Details
From the project READMEOwnerSpec MCP Server
The official Model Context Protocol server for OwnerSpec.com, an independent, numbers-first reference for the equipment people own and maintain at home, in three categories: home water treatment (softeners, sediment and carbon filters, reverse osmosis, UV, iron, hardness, sulfur smell, well water testing), home plumbing (water heaters, leaks, freeze protection, toilets, faucets, sump pumps, dehumidifiers) and home air (air purifiers and CADR, humidifiers, furnace filters and MERV). Every figure on the site is cited to EPA, NSF/ANSI, WQA, DOE, AHAM, a state extension service or the manufacturer's own document, and every page carries the date its facts were verified.
Live endpoint: https://ownerspec.com/mcp
Stateless Streamable HTTP transport (JSON-RPC 2.0 over POST), no authentication, no sessions. Any MCP client can connect and start calling tools immediately.
Connect
Claude Code
claude mcp add --transport http ownerspec https://ownerspec.com/mcp
Claude.ai / Claude Desktop: Settings, Connectors, Add custom connector,
https://ownerspec.com/mcp, no authentication.
ChatGPT: Settings, Connectors, Advanced, Developer mode, then add
https://ownerspec.com/mcp as an MCP connector (no auth).
MCP Inspector
npx @modelcontextprotocol/inspector https://ownerspec.com/mcp
Tools
| Tool | What it does |
|---|---|
search_pages | Full-text search over every page, filterable by layer (guides, calculators, parts, reviews) and topic |
get_page | Fetch one page as clean Markdown with canonical URL, dates and sources |
get_quick_answer | The page's cited 134 to 167 word quick answer, its fact sheet (answer, formula or compatibility basis, inputs, assumptions, method, typed primary source), verified date, FAQ and typed source list: the passage to quote |
read_water_report | Lab, test-kit or utility results in, each compared with the EPA limit on OwnerSpec's EPA table, with unit conversion and the guide to read |
diagnose_water_problem | Symptom or lab result in, the matching diagnosis guides and their quick answers out |
troubleshoot_home_equipment | A beeping alarm, a popping water heater, a pump that will not start: the guides that explain it |
find_replacement_part | Model or part number in, the cartridges, lamps or parts that fit out, with the cross-reference page |
get_product_picks | The picks from the review and parts pages that match a need, with the review page that compares them |
list_calculators | Every OwnerSpec calculator with its id and input fields |
run_calculator | One calculator prefilled with the user's numbers: the page's live calculator in apps, the formula in text |
convert_water_hardness | gpg, mg/L, dH, fH, Clark and mmol/L, with the USGS band |
size_water_softener | Grain capacity between regenerations, nearest nominal size and the salt-efficient Fleck 5600SXT row |
plan_home_maintenance | A dated maintenance schedule, each interval quoted from the maker or body that prints it |
All tools are read-only and annotated readOnlyHint: true, destructiveHint: false.
MCP App
In hosts that support MCP Apps
(io.modelcontextprotocol/ui), eight tools render an interactive view: the live calculator
with the user's numbers, a flagged water report table, a dated maintenance plan, and cards for
guides, parts and picks. The view (functions/_lib/widget.js, served as
ui://ownerspec/view-v1.html) is one self-contained HTML document with a hand-written
postMessage bridge, no SDK and no third-party origin. Calculators run the site page's own
form and script, taken from /index.json, so the in-chat answer is the page's answer.
Every answer ends with the URL to cite and the facts-verified date. The two ported calculator
tools use the same constants as the calculator pages on the site, run_calculator runs the
page's own code, and the site's build fails if either side drifts.
The server also exposes resources (llms.txt, llms-full.txt, the sitemap, the search
index, the OpenAPI description, auth.md) and one prompt, answer_from_ownerspec.
A2A agent
functions/a2a.js is a minimal A2A agent at
https://ownerspec.com/a2a: a question in a SendMessage (or message/send) call, a
completed task out whose artifact holds the matching pages, quick answers and URLs to cite.
Card: https://ownerspec.com/.well-known/agent-card.json.
Markdown for agents
The whole site is agent-readable, not just this server:
- Any page returns Markdown when requested with
Accept: text/markdown, or at its URL plusindex.md. - Discovery files:
/.well-known/mcp/server-card.json,/llms.txt,/.well-known/agent-skills/index.json,/.well-known/api-catalog,/.well-known/ard.json,/openapi.json,/auth.md. - Human documentation: ownerspec.com/mcp-server.
Transparency
OwnerSpec is an affiliate-funded site, but this server returns no product links, no prices and no ads: picks come back with the OwnerSpec review page that compares them, and the affiliate links and their disclosure live on that page. Editorial content is free to read, with no paywall and no auth. The site makes no medical claims and neither do the tools: contaminant figures are quoted as EPA, WHO or state limits. Privacy: ownerspec.com/legal/privacy (section 2.5).
Self-hosting
This is a Cloudflare Pages Function (functions/mcp.js, functions/a2a.js, shared code in
functions/_lib/). To run your own instance:
- Deploy the
functions/directory with a Cloudflare Pages project in front of a static site. - The server reads its content live from the site it fronts: a build that emits
/index.json(one record per page: title, url, layer, topic, dates, quick answer, FAQ, sources, picks, text) and a Markdown twin at<page>/index.md. Point theSITEconstant at your deployment. - Replace the calculator constants in
mcp.jswith the ones printed on your own pages, or remove those two tools.
This repository mirrors the production functions. Keep them in sync with your own site config if you fork it.
License
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.