The API is organised as level, source and topic, so the first job is always working out which combination holds what you want — that is what the endpoint resource is for. Once you know it, ask for an aggregate rather than records: totals by state come back computed instead of arriving as raw enrollment lines you then have to add up.
A server over the Urban Institute's Education Data API, which publishes US education data organised by level, source and topic. It gives you one tool for detailed records and one for aggregates computed by the API itself, so you do not pull rows in order to sum them.
- Detailed records for a chosen `level`, `source` and `topic` — for example schools, ccd, enrollment — with optional grouping, filters, variable labels and a result limit — `get_education_data`
- Aggregates computed API-side: pick a statistic such as sum, avg, count or median, the variable to summarise, and what to group by — `get_education_data_summary`
- Filters expressed as values, including lists, so a query can scope to specific years, states and grades in one call
- Endpoint browsing through the `edu-data://endpoints/{level}/{source}/{topic}` resource
No account and no key — the Education Data API is public. Node dependencies installed and built, with the clone linked so the client can reach it. The package.json names it `edu-data-mcp-server` 0.1.0, which npm does not have.
One command — npx -y github:ckz/edu_data_mcp_server