Docs / MCP server / MCP server
MCP server
A free, read-only Model Context Protocol server. It gives an AI assistant 10 prevalence entries from 7 population studies and ranges, 93 references, and the data behind the statistics page's charts, so a student or journalist can ask "what share of children have dyscalculia, and which study says so?" and get an answer with its source. Hosted at https://mcp.dyscalculia.io/mcp, or run locally with npx -y dyscalculia-mcp.
Hosted or local
| Hosted | Local (npx) | |
|---|---|---|
| Install | Nothing | Node.js 18+, npx -y dyscalculia-mcp |
| Works in ChatGPT and Claude custom connectors | Yes | No (they cannot start a local process) |
| Works offline | No | Yes |
| Tools | Same 6 | Same 6 |
Connect your client
- Claude Code: Add the hosted URL with one command. A local install works too.
- Claude (claude.ai and Desktop): A remote server is added as a custom connector. It needs the hosted URL; npx does not work here.
- Claude Desktop (local): To run the server on your own machine with npx, add it to Claude Desktop's config file.
- ChatGPT: Only a remote (hosted) server works. ChatGPT cannot start a local npx process.
- Codex: The Codex CLI and the IDE extension read the same config.toml. Use --url for hosted, npx for local.
- Cursor: Add a URL or a command to mcp.json, for one project or for every project.
- VS Code: The workspace file is .vscode/mcp.json. Its top-level key is servers, not mcpServers.
- Gemini CLI: Use gemini mcp add or edit settings.json. The key for a remote server is httpUrl.
- Any other client: Any MCP client that supports Streamable HTTP works with the hosted URL. Clients that only support stdio use npx.
Tools
| Tool | What it returns |
|---|---|
get_prevalence | Prevalence figures with sample size, publication year, age, reference key and a calc flag; filter by study, age or percent range. |
list_studies | Every source the statistics page cites, with the topics it supports (prevalence, sex ratio, comorbidity, persistence, family, maths anxiety, interventions, ...). |
get_reference | One full citation by key, with DOI or PMID where the site has one. |
explain_screening_test | The five screening games: what each measures, item counts, how items and games are scored in general terms, optional item sets and the limits. A description, never a result. |
list_chart_data | The 4 charts and 3 tables of the statistics page with their CSV, JSON and SVG addresses; with an id, the rows and reference keys. |
get_citation | License, citation text, an HTML credit link and the download addresses. |
Parameters and real example responses are on the tool reference.
What it does not do
- It does not screen or assess anyone. It returns published figures and a description of how the free screening test works. It takes no answers, returns no scores and gives no result. The test itself runs on dyscalculia.io and is a screening, not a diagnosis; only a qualified professional can diagnose dyscalculia.
- It does not invent a single world rate. No published meta-analysis gives one pooled figure, so the server returns the range from population studies and each study separately, each with its year.
- It gives no advice about a child. The figures describe groups of children, not one child.
- It writes nothing, opens no account and spends no credits. All six tools are read-only and need no API key or sign-in.
Example questions
- “How common is dyscalculia in school-age children, and where does that number come from?”
- “Which studies on dyscalculia and dyslexia overlap are listed, and how big were they?”
- “I am writing a school report: give me the prevalence figures with a citation for each.”
- “How does the free screening test work, and what does it not tell me?”
- “Give me the data behind the dyscalculia charts as a table.”